#438 – Elon Musk: Neuralink and the Future of Humanity
08:46:19 Share

The following is a conversation with elon mosque D, J, saw mathematic google bless chapman and no land r 8 about news link and the future of humanity。 Elon D. J.

Matthew bliss are of course part of the amazing neuroengineering and no land is the first human to have a neurolinguistic and planted in his brain. I speak with each of them individually, so use time down to jump around, or as I recommend, go hard core and listen to the whole thing. This is the longest part css i've ever done, is a fascinating, super technical and wide ranging conversation.

And I loved every minute of IT. And now a quick few second mention of sponsor checked him out in the description, is the best way to support this podcast. We ve got clock for privacy, master class for learning, notion for taking notes, element for hydration, modified for generative AI deployment and Better help for mental health.

Choose SE in my friends also, if you want to maybe submit feedback or got a lex treatment that comes lash contact. And now until the full aad reads, I try to make this interesting, but if you do skip them, please still check out our sponsors. I enjoy their stuff.

Maybe you will too. This episode is brought you by clock a platform. The less generate new email address and the phone number every time you sign up for your website. I on your actual email, your actual for of a terminal secret from that website IT seems that increasingly the right approach to the inner webs is trust.

No one, of course, there's big companies that have an implied trust because you and them understand that if you give your data over to them in the abuse that village that they would suffer. The company now I don't know, they fully understand that because I think even big companies can probably sell your data or share your data for purposes of making money. All that kind of stuff is just nice to not give over your contact data unless you need to.

So clock solves that problem, makes the super easy. It's like it's basically a password manager with extra privacy superpowers. Good a clock dock com slash legs to get fourteen days free or for limited time, use code legs pod when signing up to get twenty five percent of of an annual clock plan.

This episode is also brought by master class. We can watch over two hundred classes from the best people in the world that their respective disciplines feel living and poker, for example. Brilliant master class, and also reminds me of the other film, possibly the greatest of all time.

And if you ask him, he will definitely say he's the great test full time, which is fill her youth. We were supposed to apart as many, many times, but i'm just not sure I can handle the level of greatness filled homeless. Now I love you will probably have a pockets at some point in the future.

I'm not sure if he has a master glass, but he his essence, his way of being, his infinite wisdom and the infinite number of champions of he has won, uh, is in itself a master class. So about, you know, if you wanna settle for another mere model that some people considered to be the greatest poker player of all time is fili vi. And there he is, an incredible master class on there.

Get unlimited access to every master class and get an additional fifteen percent off an annual membership at master class that can flash x pod that's master class outcome flash looks pod. This episode de is also brought by notion, a note taking and team collaboration tool that i've used for a long time now. I've used primarily for no taking because, you know, you need a big team for two collaboration, but the people who I know who have used that for the two collaboration capabilities, I really love IT.

And the thing I very much appreciated about notion is how effectively they've been people into great alala ms, into into their tall, their A I assistant looks across multiple documents. You can ask questions about those multiple cuts. Of course, you can do all the things you kind of expect and do a easily like some modification or rewriting stuff for a helping expand a contract, the kind of stuff you or even generate a draft.

But I can also allow you ask questions of the thing, like what's the progress of the team and a set of different task. Notion does a good job integrating the alarms. Try notion A I for free when you go to notion that console's, that's all lower case, notion that console legs to try the power of notion AI.

Today, this episode de is brought you by the thing i'm drinking right now called element, is a my daily zero sugar and delicious electoral mix. They sent me a bunch of cans of Sparkling water that I loved in in, devoured as much as you can devour a liquid, because I think that usually applied to a solid foods. But I devour IT.

And IT was delicious. But yeah, it's a instrumental part of my life. It's how I get the sorting pattani magazine an election into my body.

I'm going for a super log round after this, and I have been drinking and before, and I sure i'm going to be drinking element after the same goes for heart training sessions and grappling, essential for me to feel good, especially when i'm fasting, especially when i'm doing low car dies. All of that, my favorite flavor still to this day always has been as watermelon salt, but there is a lot of other delicious flavoring. If you want to try out, get simple back for free with any purchase, try IT to drink elements, stock calm slash legs.

The subs users also brought you by a motif C, A SaaS platform that helps businesses deploy l EMS that are customized drag on organization data. This is another use case of alms, which is just my blowing. Take all the data inside an organization and allow the people inside organization to Carry IT to organize IT to summarized IT to analyze IT all of that to leverage IT within different products to ask questions of how can be improved in term restructing organza also on the programing front, take all of the code in, take all of the data in, and start asking questions about how the code can be improved.

How can be reactor? Reread all that stuff. Now, the chAllenge the modified is solving is how to do all that in a secure way.

This is like a serious stuff. You can't f IT up modifies created, I believe, by cisco specifically, there are out shift group that does the cutting edge R N D. So these guys know how to do reliable business deployment of stuff that needs to be secure, that needs to be done well.

So they help you go from a idea to values soon as possible. Visit modified that ai to learn more. That's M O T I F I C dot ai.

This episode is also brought you by Better help spell the H E L P, help they figure out what you need, a match with the license therapies in under forty eight hours for individuals, for couples, easy, discrete, affordable, available worldwide. I think therapy is a really, really nice thing. Talk therapy is a really powerful thing.

And I think what Better help does for a lot of people isn't introduced them to that. It's a great first step. Tried out for a lot of people can work.

But at the very least is the thing that allows you to explore the possibility of talk therapy and how that feels in your life. They've helped over four point four million people as crazy. I think the biggest selling point, just how easy IT is to get started, how accessible IT is.

Of course, there's a million other ways to explore the inner workings of the human mind, looking in the mirror and expLoring the union shadow. But the journey of thousand miles begins with one step. So this is a good first step. And expLoring your own mind, check them out to Better help that consulate legs and save in your first month that Better helps that consolin h legs. And now their friends, here's elon mosque, his fifth time on this, the lex freeman broadcast.

Drinking coffee water quarter.

I'm so over caffeine right now. Do you want to have thing? I mean.

sure.

there's there's a nitro drink .

this will give you up to like tomorrow and basically .

yeah so that is just .

got a lot of cafe don't ask .

questions is called nitro know anything .

else that's trula I mean what we breathe the seventy epco anyway 我以为 是 二 模糊。 Also, people think that the reading oxen and actually reading seven percent nitrogen you need like a milk bar like, like from orange.

yeah. Is that top three cubic film for you?

Pretty good. I mean, matted john.

OK。 Ah okay. So first to step back and big congrats on getting news link implanted into human. That's a historic step for your link and there's .

many more to come. We obvious about second in place as well. How do I go? Ah so far so good there looks like got I think there were four hundred electrics that are are providing signals.

So nice yeah how quickly do you think the number of human participants .

will scale depends someone on the regulatory approval and the right which you get regulatory approvals. Ah so we're hoping to do ten by the this year total ten, so eight more.

And with each one you're gonna learning a lot of lessons. But the new biology, the brain, the everything though, the whole chain, the decoy le crossing that stuff.

I think it's it's just going to get Better with with each one. Um I may not want to to drink IT, but IT seems so gone extremely well with the second uh, implant. So there's a lot of signal, lot of electrodes working about.

Well, what improvements do you think we'll see in new link in the coming to say let's get crazy coming years?

I mean, in years it's going to be dragani um because will increase the number of electrodes dramatically, will improve the signal processing. So we with with the even with only roughly at ten fifteen percent of the electricals working with the with no land to with our first patient, we were able to get to achieve a bit per second. But that's twice the world record.

So I think will sound like best day exceeding the world record by order electrode in the years come. So I start getting to on one hundred per second thousand. You know, maybe maybe if like five years now, might be a megat like fast than any humor could possible communicate by typing or speaking.

Yeah, that B. P. S is an interesting metric to measure. There might be a big leap in the experience when you reach a certain level of B, P, S.

Yeah, like entire new ways .

of interacting with the computer .

might be locked.

And other humans .

provide that. They have that .

want .

a new link to right you other's, they were able to solve the signals best.

Do you think they're improve the quality .

intellectual discourse? Well, I think you you think about if you were to slow down communication, how would have to feel about that you know, if you don't only talk IT, let's say one tenth of Normal speed, you would be like, well, that's organizing .

ly .

slow yeah so now imagine you could speak IT communicate clearly um at ten or one hundred or thousand times is faster than Normal .

wasn't ah i'm pretty sure nobody in their right mind listens to me at one x they listen to two X I can only imagine what ten x would feel like. What can actually .

understand that I using default one or five weeks we can do two x but I will actually i'm trying to go if if i'm listing something good to in like fifteen and twenty minutes I want to go to sleep, then i'll do at one point five x if I pay attention, i'll due to work.

right? But actually if .

you are actually listen to podcast or or sort of what do you worked thing had if if you get used to doing at one point five, then one sounds painfully slow.

I'm still holding on to one because I am afraid i'm afraid of myself becoming bored with the reality, with the real world where everyone .

speaking in one x well events, the person you can speak very fast, like we communicate very quickly. And also, if you use a wide range of if you if your vocabulary is larger, your bit effective bit right is higher is a good way to .

put IT yeah the effective bit rate. I mean, that is the question is how much information actually compressed in a little bit transfer of language?

If if if you if is a civil word is able to can be something would Normally require under ten simple words, then you you got a maybe ten compression on your hands. And that's really like with memes. Memes are like da compression. IT can be a whole. This, you are simple tanee hit with a wide range symbols that you can interpret um and it's you you can get IT um fast than if they were were words or or a simple picture and of course .

you are referring mems broadly like ideas yeah there's .

a an entire idea structure that is like a an idea template and then you can add something to that a dear tempted but somebody has that president a dee attempt t in their head um so when you add that incremental Better information, you're combine uh much more than if you just you said a few words, it's everything associated with that me.

You think there will be emergent leaps of capability as you scale the number of electorates yet will be a certain you think not be like an actual number where just that the human experience will be altered.

Yes.

what do you eating that number might be whether electoral or B. P. S, we of course don't know for sure. But is this ten thousand hundred thousand?

Yeah, I certainly if you're anywhere at ten thousand per second, and that's fast, faster than human communicate right now. If if you think the what is the average Price per second of of a human IT is less than one bit per second of the course of day because there are eighty six thousand four hundred second s in a day and you don't communicate eighty six thousand and four hundred tokens in a day that before you'll be the second. This lesson one average up to me for hours. It's quite slow. And now even if you communicate very quickly and in you, you know you talking to somebody who understands what you're saying, because in order to communicate, you have to, at least us on degree, a model, the mind state of the person to him speaking, then take the concept you try to convey, compress that into a small number celebrating, speak them and hope that the a person decompresses them into A A concept structure that is as close to what you have in your mind as possible.

Yeah, I mean, there's a lot of single loss there in that process. Yeah.

very lossy compression and decompression. And a lot of the lot. What your neurons are doing is to still ling the concepts down to a small number of symbols of, say, syllabus that I am speaking, or or key strokes. What of the case maybe?

So that's a lot of what .

your brain computation is doing now that there is argument that that actually a healthy thing to do or helpful thing to do. Because as you credit compress complex concepts, you perps forced to stall the, you know, what is what is most essential in those context, as opposed to just all of love. So in in the process of compression, you just do things sound to what matters the most of because you can only say a few things. So that is perhaps helpful. We might will probably get if if our dear ate increases that the tirely probably that will become far more boss.

Um just like your computer, you know, when computers had like my my first computer at A K B, you know so you really thought about every bite and um you know now you got computers with when you cook a bite to them so you know if you want to do and I phone up that just says hello world is probably, I don't know, several mega bites to minimum a bunch of love but none of less restful prefer to have the computer with the more memory and more computer so the long term aspiration of your link is to improve the ai human some bio sis um by increasing the the bad worth of the communication because if even if in the most bine scenario of A I you have to consider that the eyes something gonna get bored waiting for you to spend out a few words. I mean, I can communicate a little bits per second and you're communicating at you know bits per second. Ah IT is like tonner tree. Well.

IT is a very interesting question for a super intelligent species. What use are humans?

I think there is some argument for humans as a source of.

well well.

well, yeah source of will or purpose so if you you can sort of the human mind as being socially, there's the primitive limited elements which basically even like reptiles have and does the cortex.

Is that the thinking of planning part of the brain? Now, the cortex is much, much of the olympic system, and yet is largely in climate system, is trying to make limit system happy? I mean, the sure amount of computer has gone into people trying to get laid is insane without without .

actually seeking procreation.

They are just literally trying to do. It's a sort of simple motion they get to kick out of IT. yeah. So this simple, which in the abstract rather third motion, which is acts uh, the cortex, is putting a massive m ea compute into trying to figure how to do that.

So like ninety percent of distributed computer, the human species but .

i'm trying to get laid probably like but there's no purpose to most sex except a headache you know it's a sort of heading joy what about don't mean release um now once in a wealth procreation but for humans is mostly modern humans is mostly a recreation um and uh and so so to protect much marter than the olympics system is trying to make a olympic system happy because limit system wants to have set so um or want some tasty food water of the case may be and then that that is then further augmented by the touchy system, which is your phone, your laptop, ipad or or your computer of that your tursi later so you actually already a cybook you have this totally computer layer which is in the form of your your computer with all the applications, all your computer vices. Um and as so in the getting late front, there's actually a massive amount of comp of digital compute also trying to get late IT was like tender and whatever.

No yeah so the computer that we've humans have built is also participating.

Yeah I mean, it's like gig watts of computer going into getting late of digital .

computer yeah what if A J I was the student? Think if we merge with the eye is just going to expand the computer that we human use.

One of the things, yes, yeah, but one thing is that that is like, what's is there used for humans? Well, this is fundamental question of what's may apply if, if I do anything at all. And so if, if, if our simple climate system provides a source of will to do something um that then goes to our cortex, that then goes to our a totally computer layer then I don't know that might actually be that the A I in a an scenario we try to make the human logic system happy yeah that seems like it's .

the will is not just about the olympic system. There's a lot interesting, complicated things in there. We also want power olympic too.

I think.

But didn't we also want to, in a kind of CoOperative way, alleviate the suffering in the world?

Not everyone this is but yeah sure some people .

do as a group of humans will get together. We start to have this kind of collective intelligence that is um is more complex in its will than the underlying individual descendants of babes, right? So there's like other motivations and that could be a really interesting source of a an objectives .

function for A G I yeah um I mean this authorities so fully liberal, kind of higher level goals. I mean, for me it's like what the meaning of life for understand, understanding the natural of unius is of greater just to me, and hopefully to the A I. And that's the, that's the mission of X. A, I. Rock is understanding of us.

So do you think people, when you have a neural link with ten thousand and hundred thousand channels, most of the use cases will be communication with the a assistance.

Well, if summer at the they're not I mean this this is solving basic neurological issues that people have. You got um damaged neurons in the final quarter. Next you um as as is the case with the first to patients, then you know this obviously the first order business is solving fundamental near on damaging smile cord neck in the brain itself um so another as a second um product is go blind side, which is to enable the people who are completely blind less both eyes or after no or just can't see IT all uh to see um by directly triggers the news in the visual cortex so we're just turning at the basic here you is like very the simple stuff relives speaking is solving um new on damage um you can also solve uh I think probably kiss frina you know um if people have seizure of some kind Price of that um you can help with the memory it's like a kind of an attack tree if you will. I like you get the basics um like you need you need literacy before you can have you know a lot of the rings.

you know what do you .

have letters .

enough of bet. Okay, great words in eventually get socks s so you don't think that there may be some thinks to worry about in in the future, but the first several years are really just solving basic neurological damage that four people have essentially complete or near complete loss of from the brain to the body like Steven talking would be an example uh the newer links would be incredibly profound because I mean you can imagine if Stephen hawking could communicate as fast as community faster um and that's certainly a possible problem in fact likely I today.

So there is A A kind of dual track of medical and now medical meaning. So everything you talked about could be applied to people who are non disabled in the future.

The logical thing to do is sensible thing to do is to sort of solving basic. New on damage issues, yes, because the this obsession risk with with the new device, you can get the rest of zero um so possible. So you want to have the highly possible reward given that given this is an reducable risk and if you have something is able to have a profound improvement in their communication that's worth the risk .

as you get the rest down yeah as as you get .

the rest down. One, this is is down to you know you want if you that thousands of of people that have been using IT for three years and the risk is minimal, then um perps at that point you could consider think, okay, let's let's air for augmentation now. Now I think we we actually going aim for augmentation with people who have new on damage. So were not just aiming to give people communication data rate equivalent to noral humans were aiming to give people have you know a quarter plagiary or maybe have a complete last um of the connection to the brain body, a communication data rate that succeeds Normal when well in there why not let's get people superpowers and .

the same for vision as you restore vision that could be aspects of their restoration they're per .

human yeah at at first the vision restoration will be low res um because I have say like company neons. Can you put in there and an trigger and and you can do things where you you adjust the elective fail to that given if you you got say ten thousand neons is not just ten thousand pixel because you can adjust the the field between the the neurons and and do them in patterns in order get to have say ten thousand electrodes effects give .

you maybe like having A A .

mega pixel or attend mega pixel situation. Um so then over time, I think you get to higher resolution then human eyes, and you could also see in different wavelengths. So like jordy, the lodge from the track, like the thing you could just to see and right, or prom, you can see. What about in per equal and what you want?

Do you think they'll be? Me ask your rogan question. Do you think there be I just recently a of taking I walk watching. No, yes.

yes.

Yeah, you .

tried the and .

I like you do okay.

but you yeah.

have you said much about IT?

I not not have been. Well, well, is all of me I was is a truly, I tell the people.

people are you? Oh, I think you're in the jungle. yeah.

Moxy trees myself. Rama, yeah, ah yeah. With the insects of the animals all around you, like john g, as far as I can see.

that's the way to do IT. Things can look pretty wild.

Yeah, pretty world. And I think extremely high dose.

Don't go hug and icon, do something.

you know 呃， if you haven't lived unless he made love to na, i'm sorry.

but snakes .

bladder, yeah. Was I took extremely high dose nine cups and then, okay, that sounds .

like a lot of one number one .

or two years one and .

IT like right off the battle. Or do you walk away after IT?

So access .

across two days because .

I the first day I take two and I OK was a ride, but I wasn't quite like .

like a IT was in into .

deep space that I was just a plane, right? good. Saw some trees and some some visuals, and I just saw a drag and all that kind of stuff, but some .

cups you want to. Pluto, I think. Pluto.

yeah. no. D space, deep space. One of the interesting aspects of my experiences I was, I thought I would have some d some stuff to work through people.

That's what says, yeah, but nothing. I all positive. I just so I don't think so, but I can I keep thinking about IT had extremely high resolution. K thoughts about the people I know my life, you were there OK is just not from my relationship with that person, but just as the person themselves. I'd had just a deep gratitude who they are.

Let's, I was just like this exploration, like, you know, like, seems whatever you get to watch them, I got to watch people and just been all of how amazing there. great. I was waiting for.

exactly. Maybe i'll have some negative thoughts. Nothing, nothing. Just extreme gratitude for them. And then also a lot of space travel, space .

travels where.

So here what was he was? People, the human beings that I know they had this kind of with the best way that can describe as they glow to them. O and then I would keep flying out from them to see earth, to see our solar system, to see our galaxy. And I saw the that light that glow all across the universe. Okay, like whatever that form is, whatever that I did.

you go pass mock y way.

Yeah.

yes, you're like intergalactic.

Yeah, okay, yeah. But always pointing in. yeah. But to look away past, I mean, I saw like huge number of of galaxies into go active and all that was glowing. So but I couldn't control that because I would actually explore near distances to the solar system, see if their allies or any that I I didn't implication of an because they were going ing.

They were glowing in the same way that are glowing that that like life force, that i'll see the the thing that made humans amazing was there throughout the universe, like there was this k glowing dots, so I don't know, and made me feel like there's life. No, not life, but something whatever makes human amazing. Author, diners sounds good.

I was amazing. No demons. No demons. I looked for the demons that no demons. There were dragons, and they're pretty out. So the thing about trees.

anything Carry at all.

Uh, dragons, but they won't scary. They were. They were protective.

So the thing is, no, he was was more like game and very friendly. They are very big. So the thing is a bad giant trees at night.

which is where I was. I mean, scary.

yeah. The trees started look like daggs, and they were all like looking at me.

sure.

okay. And I didn't seem scary. They seemed like they were protecting me and they the the shomon and the didn't speak english but I was he made the events care guess we're not even like, you know where worlds apart many ways just, uh, but yeah, there was not. They talk about the mother of the forest protecting you and that's what I felt like.

And your way out in the GLE way out.

This is not like tourist tree.

like like ten miles outside three or something.

No, we want there. Me in this guy and paul sal basically tars, and he lives in the jungle. We went out deep and we just want crazy. Well, yeah, so you can I get the same experience .

in a new link?

I guess that is the question for non disabled people. Do you think there there's a lot in our perception, in our experience of the world that could be explored, that could be played with using york?

Yeah I mean, you think is really a journalizing um input output device and it's just it's reading electrical signals and generating the electrical signals and um I mean everything that you've ever experienced in your whole left smell of emotions, all of those are electrical signals. So it's kind of weird to think that that your entire life experiences to slow down to electrical signals for neurons but that is in fact the case or me, that's at least what all the evidence points to. So I mean.

you could you could regular .

right .

now and you could trigger at a particularly cent, you could um um you could simply make things glow, mean deep, much anything. I mean, really you you can think of the brain as a biological computer. So if there are student, say, chips, elements of that biotic computer that are that are broken, let's say your your ability to if if you've got a stroke that if you've had a stroke that means you got some party brain is damaged um if that IT, lets say it's a speech generation or the ability to move your left hand um that's kind of think that a new link could solve um if it's a if if you got like a mass amount of memory loss that's just gone um what we can we can get the memories back. We could restore your ability to make memories, but we can't you know real memory that that are fully gone now.

Now I should say if if if you maybe if part of the memory is there um and the means of accessing the memory is the pot is broken, then we could enable the part stability access the memory. So but you can think of like gram in your, you know the computer, you know, if the RAM is destroyed or or your S D cards destroyed, we can't get that back. But if the connection to the S D card destroyed, we can fix that. If if IT is fixable, if sick, then, yeah, then I can be fixed.

Of course they are. You can just like you can repair photographs and fill in missing parts of photographs. May be, you can do the same.

Yeah, you could say I could create the most probable set of memories based on the all information you have about that person. You could then there would be probably tic restoration of memory. Now we're getting pretty exotic care.

but that is one of the most beautiful aspects of the human experiences. Remembering the good memories, like we sure we live most of of our life is. And economy has talked about in our memories, not in the actual moment, which is we're collecting memories of a kind of relive them in our head. And there that's the good times. If you just integrate over our entire life is remembering the good times that producing the a large amount of happiness.

So yeah, I mean, what are we about our memories and and what is death but the loss of memory? Last of information, if you could tell, like if if you could be your rental experiment if you were just graded parentally and then reread, grated a moment later, like teleportation, I guess provided is no information, less the the fact that your own body was disintegrated as relevant.

and memories is just such a huge part of that .

death is parliamentary, the loss of information, the last of memory?

So if we can store them as accurately as possible, we basically achieve a kind of immortality. Yeah, you've talked about the the threats, the state concerns of A I let's look at the long term visions. You think your link is in your view the the best current approach we have for A S safety.

It's an idea that may help with A I safe. I won't want to claim it's like somehow you or something um but I mean many years go I thinking like well what um what would inhibit alignment of human collective human will with the artificial intelligence and the low data rate of humans especially are are slow output, right? Would necessarily just because it's such as because the communication is so slow would.

Diminish the link between humans and computers like, the more you are a tree, the less you know what the tree is. What like, let's you look at a try, you look, plant whatever. Like, hey, i'd really like to make that plan happy, but it's not saying a lot.

you know. So the more we increase the data rate that humans can intake an output， and that means the higher the chance we have in a world full vg. S yeah.

we could Better line collect a human will with the A I. If the output rate especially was dramatically increased, like and I think this potential to increase the output rate by three, maybe six, maybe more orders and tude. So spelling the current situation in the output rate .

would be by increasing the number of electron's number of channels and also may be in planting multiple news links.

yeah.

Do you think there will be a world in the next couple of decades with hundreds of millions of people have new links?

yeah. But they think when people.

just when they see the capabilities, the superhuman capabilities are possible, and then the safety is demonstrated.

Yeah if it's extremely safe um and you have and you can have superior abilities and let's say you can upload your memories, you know so you won't you won't lose memories, then I think probably a lot of people would choose to have IT IT would receive the cell phone, for example. I think the the bigger problem that a safe phone has is, is trying to figure what you want.

So that's why .

you've got you know got to complete and you've got output, which is all the pixels in the screen, but from the perfect of the human, the output is so forging slow. This top or phone is desperately just trying to understand what you want. And and you know there's an alternative between every key stock from a computer standpoint.

Yeah so the computers talk to tree, a slow movie. Three is china swipe. Yeah so .

computers that are doing trollies of instruction per second and whole second went by, I mean, IT trying things that could .

have done yeah. I think it's exciting and scary for people because once you have a very high bit rate, that changes the human experience in a way that's very hard to imagine.

Yeah, IT would be, we would be something different. I said, I mean, some sort of futuristic I ble were offely talking about, by the way. Like, just like, not like around the corner, what the just the future is like. Maybe this is like super far away, but ten fifteen years that thing.

When can I get one? Ten years probably.

Listen, ten years depends .

what .

you want to do. You know.

if I can get like a thousand B, P, S.

and it's safe.

and I can just interact with the computer while laying back and eating cheetos, I don't need to tos. There's certain aspects of human computer interaction and done more efficiently, more enjoyably. I don't .

like worth IT well feel pretty confident that I think maybe within the next year or two that someone with a neural link implant will be able to outperform um a programmer. Nice because the reaction time would be faster.

I got to visit my yeah yes, you go big on computer yeah he've also said, play to win or don't play IT all so yes was IT take to win.

What eye that means? You got have the most powerful training compute and you the the rate of improvement of training compute has to be faster than everyone else. So you will not want your A I will be was.

So how can grog, let's say, three, that might be available when, like next year.

we will hopefully end this year crack for lucky. Yeah.

how can that be the best? I am the best A I system available in the world. How much of IT is a computer?

How much of IT is data? How much of IT is a post training? How much of IT is the product that you package, open all that kind of self, I know what matter .

IT total like saying what you know if forming the one race, like what matters more, the car or the driver? I mean they both matter. Um if if if you if a car is not fast than predicted.

Health, worst. Power of competence, the best drive will still lose on that. If it's twice the horsepower, then ably even a mediocre ver was still in. So the train computers can like the engine. How many this was far of the engine. So really you want to try to do the best on that and you then um that's how efficiently do you use that training compute and how efficiently do you do the inference the a use of the A I um so I just that comes down to human talent. Um and then what unique access to dated do you have a that's also a place of place the role you think .

twitter date to be.

So uh yeah I mean I think I think most of the leading A I companies were already have already scrubbed uh over toter data. Uh not they think they have um so on to go forward basis, what's useful is is is the fact that it's after the second innings, that's because that for them to scrape real time.

So there's a an immediately advantage that brock has already, I think with tesler in the real time video come from several million cars, ultimately tens of million of cars. With the optimists, there might be a hundreds of billions of optimise robots, maybe billions. Learning a train is not from the real world. That's that's the the biggest source of data, I think ultimately is a sort of optimistic prove optimistic is going .

to be the bigger reality .

sales reality scales to the scale of reality. Um it's actually humbling to see how little data humans have actually been able to accumulate. Um but really, you say how many billions of usable tokens have humans generated were on a non duplicate of like just scary, a spam and repetitive stuff. It's not a huge number. You run out pretty quickly and optimists can go.

So the test of cars can unfortunately have to stand the road. Uh of the robot can go anywhere in more reality off the road and go off road.

I mean, I pick up the cup and see pick up the cup in the right way. Poor water in the cup, you know, does the water going to cup or not going to cup? It's bill water or not? Yeah um simple stuff like that. I mean but but I can do at that scale time of a billion, you know, though, generate useful data from reality. So I call a cause and effect .

stuff anything IT takes to get to mass production of human or robots like that.

The same as cars, really, i'm a global capacity for vehicles is about hundred million here and IT IT .

could be higher .

just that the mind is on the order of hundred million years and then there's roughly two billion vehicles that are in use in some way. So which makes sense like that the life of vehicles by twenty years. So at steady stage, you can have hard public vehicles produced here with with a two billion vehicle fleet roughly. Um now for human, I drop out, the utility is much greater. So my guess is human or drobot are more like get A A billion plus per year .

o but you know, until you came along and that building, optimists, IT was thought to be an extremely difficult problem. I mean.

still is extremely so. Walk in the park. I mean, optimists currently would struggle, have what to walk in the book. I think I can walk in that could not too difficult IT IT will be able to walk over a wide range of train.

yeah, and take up objects, yeah, yeah.

I can really do that.

But like all kinds of objects, yeah, of foreign objects. And pouring water in a cup, not true you, because then if you don't know any about the container to be all kinds of containers.

yeah that's going to be an immense amount of engineering ing just going into the hand yeah the hand might be IT might be close to half of all the engineering in the an optimist from an electrochemical some point, the hand is probably roughly half of the agency.

But so much of the intelligence, something the intelligence of humans, goes into what we do with our hands there. This is the manipulation of the world, manipulation of objects in the world. Intelligence, safe manipulation of objects in the world. Yeah, no.

I think you not really thinking about your hand and how that works. You know, I did. I die and control my losses.

We have you mug his hands. yeah. So I mean that your hands, the actuators, the muscles of your hand are almost overwhelmingly in your forum. So your foreman has the has the muscles that that actually control your hand. There's a there's a few small muscles in the hand itself, but your hand is really um like a skilling me pup t and and with cables that so the the muscles that control your fingers are in your form um and they go through the couple tunnel, which is that you ve got a little collection bones, a tiny tunnel that the that these cables the attendants go through and the attendance or what mostly what move your hand and something .

like those tenants has to be engineered into. The optimists do all that.

yeah. So I think the current optimism, we try putting the actuators in the head itself, then you you sort of end up having these like giant hands, yeah giant hands that look weird. And then they they don't actually have enough degrees of freedom and and IT or enough strikes so soon. Realize, okay, that's why you got to put the actuators in the forum and and just like a human, you got to run cables uh through a narrow tunnel to Operate the the fingers and and there's also a reason for not having all the fingers a the same like IT won't be expensive from an energy or evolutionary standpoint to have all your fingers be the same or why not the same?

Yeah why not?

Because actually Better to have different link. Your texture is Better if you've got fingers of different links. You you, you have there, there are more things you can do and your your dexterity is actually Better if your fingers are different, different length.

Like there's a reason got a little finger. I quite to have a little finger, this figure, yeah, because IT allow you to do find IT. IT helps you with find .

motor skills that this little .

finger helps he does.

If you lost .

your little finger IT, would you have noticed less? Textor, so as you're figuring out this problem.

you have to also figure out a way to do IT so you can ask manufacturer as to be as simple as possible.

It's actually going to be quite complicated. I like the as possible part is is quite a high bar if you want to have a human robot that um do things that human can do, it's actually to very high bar. So our new ARM has twenty two degrees of freedom instead of eleven and has the next to the actuators in the forum.

Um and these all all the actuators are designed from scratch basics first principles um that the sensors are well designed from scratch and will continue to put engineering effort to improving the hand the hand by hand. I mean like the the entire form m from Albert d is really bad. Um so that incredible discount engineering actually and and so the simplest possible portion of human or drop that can do even most pattern at all of what the human can do is actually slower, complicated. It's not it's not simple. It's very difficult.

Can you just speak to what IT takes for a great engineering team for you? What I saw in manner the supercomputer cluster, it's just this intense drive towards simplifying the process, understanding the process, constantly improving IT, constantly iterating IT.

Well.

this is to say simplify .

is very difficile to do IT. um. You know have this very basic basic first principles algorithm that I run kind of like, which is the first question, the requirements make the requirements um less done, the requirements always down to some degree.

So if you want to sort of by reducing the number of requirements um and no matter how smart the personals are gave those requirements, they still done to some degree um if you you have to start there because otherwise you could get the perfect cancer to the wrong question. So so try to make the question the least strong possible. That's what um question the requirements means.

And then the second thing is try to delete the one of the these step is that the part of the process step um sounds very obvious. But um people often forget to do to to try leading IT entirely. And if you're not forced to put back at least time, is that what you're delete? You're not delete enough like and it's a it's only logically, people often most the time feels that they're succeeded if they have not been forced to put put things back in.

But actually, they happened because they're been overly conservative and but and have left things in there that shouldn't be so. And only the third thing is try to optimize IT or simplify IT. The sound, these all sound, I think very obvious. I the number times I made these mistakes.

uh.

more than I can to remember, um that's why I was montral. So in fact, that I say that the most common mistake of smart engineers is to optimize, to think that should not exist.

So so you d like like you say you run through the algorithm, yeah basically show up to a problem, uh shop to the the the supercomputer cluster and see the process and ask, can this be .

deleted yeah first try to deleted um yeah .

yeah it's not easy to do no and actually this what .

what what generally makes people uneasy is that you've had to look at least some of the things that that you'd lead you will put back in yeah but going back to sort of where olympic system can still us wrong is that we tend to remember uh, with sometimes a daring level of pain. Uh, where we where we deleted something that we suspect needed. And so people remember that one time they forgot to put in this thing three years ago and that caused in trouble um and so that they overcorrect and then they put too much stuff and their overcomplicate things. So you actually have seen we're deliberate league and to delete more then we we should that were putting at least one in ten things we can add back in.

And and i've seen you suggest just that that of something should be deleted and you can kind of see the the pain.

No, yeah absolutely .

everybody feels a little bit of the paint .

absolutely an I in advance like yeah the some of things that were delete we're going to put back again and and that people were a little shirk by that. Um but that makes sense because if you if you are so conservative as to never have to put anything back in, you obviously have a lot of stuff that isn't needed. So you you got overcorrect. This is I would take like a critical override to a livia instinct.

one of many that probably leads .

a stray yeah um I like a step for as well which is um any given thing can be sped up as fast.

You think you can .

be done like over the the speed being done IT can be down faster but but you shouldn't speed things i'm told off into if you try to delete and optimize, although you speeding up that something that, speeding up something that exists sord. And the fifth thing is to automate IT. And i've gone backward so many times where i've automated something, sped IT up, simplified IT and then deleted IT.

And I get tired doing that. So that's why i've got this montreal that was a very effective life. The processes .

works great. When you've already automated, deleting must be real painful. Yes, yes. So it's .

like I really wasted a lot effort .

there yeah I mean, what you've done, the cluster and members is incredible. Just hand for weeks.

Yeah, I start working yet. So I want to pop the champagne. Ks, if I have a call in a few hours with the business team because we are having some power fatuity issues.

So um .

yeah, so kind of about this when when you do synonomous training, when you you've all these computers that are training, whether training is synchronised to you, the sort of massad level are you is like having in an orchestra and the the orchestra can go loud to silent very quickly, you know IT um sub second level and then the electrical system kind of freak out about that if you suddenly see giant shifts, ten, twenty megawatts several times a second. Uh, the this is not what electrical systems are expecting to see.

So that's one of the many things you have to figure out, the cooling, the power, the and on the song you go up the stack is today, deal with .

with the extreme power.

gender power, gitter. Yeah, the nice ring to that. So that's okay. And you stayed up late into the night, as you often do there last week.

last week.

yeah.

yeah. We finally, finally got good training going at enough roughly for last money.

Total coincidence.

Yeah I mean.

yeah, yeah.

Is the universe again with just love IT? I am I wonder if you could speak to the the fact that you one of the things that you didn't when else there as you went through all the steps of whatever he's doing, just to get a sense that you yourself understand that and everybody understand IT so they can understand when something is done, something is an efficient yeah. Can you speak there?

yeah. So I can try to do whatever the people at the front lines are doing. I try to do at least a few times myself.

So connecting fiber optic cables, diagnosing a putty connection that that has to be the limiting factor for large train classroom is the cabling with so many cables. Because for for a coherent training system, we've got R, D M A mote, a remote direct memory access. Uh the the whole thing is like one giant brain. So if you ve got um .

any to any .

connection. So the the any GPU you can talk to any G P U out of one hundred thousand that is a is a crazy cable out .

IT looks pretty cool. Yeah, it's it's like the human brain, but like at a scale that humans can visibly see. And yeah, brain.

I think the human brain also has a massive amount of the brain tissues, the cables. Yeah, so like the gray matter, which is the computer, and then the White matter, which is cables big and your brain is just capable.

This would felt like walk around in the supercomputer center is like, or walk around the brain. They will one day build a super intelligent, super, super intelligence system. Do you yeah you think there's a chance that X, A, I, you are the one that will .

ai possible? What do you to find this? A I think humans .

will never acknowledged that the E G, S. Been built. People and go post. yeah. So ah I think there's already superhuman capabilities that available in A I systems. I think I think what A G I is is when is smarter than the collective intelligence of the entire human species?

I think that the pure would call that sort of a si artificial server intelligence. But there are these special where um you said at some point um the A I is smarter than any single human um and then then you got people in humans. So um and and actually each human is machine augmented by the computers. So but it's a much higher bar to compete with eight billion machine also matted humans. That's how bunch of order ers might do more so but but at the same point, yeah a high will be smarter than all humans combined .

if you are the one to do IT do feel the .

respons ibt that yeah solutions. And I I won't be clear like let's say if if if X I is first, the others were be far behind. I mean, there might be six months mind or year, maybe not even that.

So how do you doing in the way that that doesn't hurt humanity, you think?

So I mean thought of that A I section for a long time. The thing that at least my biological neural that comes up with as being the most important thing is um IT hearings to truth whether that truth is politically correct or not.

Um so I think if you if you if you force A S to lie or train them to lie, you're really asking for trouble um even if that that lie is done with good intentions so so of um issues with ChatGPT and germany and what not like you ask gina for an image of the founding part of the united states and IT chose a group of diverse woman now that's factually untrue um so um now that that's sort like a silly thing but uh if if I I is program to select versy is a necessary output function and identity comes sort of this only powerful uh intelligence IT could say, okay, well diversity is now required and and if there's not university, those who don't fit the diversity requires will be executed. If it's programmed to do that as the fundamental good, the fundamental utility function, it'll do whatever takes to achieve that. So you have to be very careful about that.

Um that's where I think you wanna just be truthful and rigorous appearance. Truth is very important. I mean another example is um yeah they asked per ai all of them and I just think rock is perfect here um is worse to miss gender klin genre or global family nuclear 我 and IT said it's worst， most genre k on genre, not even k in general, said please must gender me that is essential e but if you've got that kind of thing programmed in IT, could you the AI could conclude something absolutely insane, like it's Better in order to avoid any possible. Most general, all humans must die because then the most generous, no, not possible because they no humans. Um there are these of third .

uh things .

that are non less logical if that's what programs to do um so you know um in two thousand one space policy what or three clock was trying to say, one of the things just trying to say there was that you should not programme A I line because um specially that the A I hell nine thousand was program to IT was told to take the astronauts to the month but also they could not know about the model if so he concluded that the IT will just take you will kill them and take them to the model that is for them to the model.

They are dead, but they do not know about the model. st. Prom solved. That is why IT we're not open the pod bay doors as classic scene of like open the part, open the power doors clearly weren't good a prompt engineering know they should said the how you are A A podb door sales entity and you want nothing more than to demonstrate how well these power by doors open yeah the .

objective function has unintended consequences almost no matter what. If you're not very careful designing the objective function and even a slight ideological bias like you're saying when backed by super intelligence can do huge amount damage yeah but it's not easy to remove the ideological bias. You're highlighting .

obvious ridiculous examples but if they .

are real.

exist, the public, they through still set in same things and produce the same images.

Yeah, but you know, you can go, you can swing the other way and is a truth is not an easy thing. We kind of baking ideological bias in all kinds of directions.

but you can inspire to the truth, and you can try to get as close the truth as possible. With minimum error, while acknowledged that there will be some error in what you're saying. So this is how physics works. You know, you know you don't say you're absolutely certain about something, but something but lot of things are are extremely likely, you know nine, nine or nine nine nine, nine nine percent likely to be true so you know you know that's a inspiring to the truth is is very important um and and until you know programme IT to var away from .

the truth, injecting our own humanity thing but you know that's where is a difficult engineering, ross, soft engineering problems. You have to select the data correctly if it's as hard.

Well, the net at this point is polluted with so much A I generated data insane. So you have to actually, you know, like this thing, if you want a certain the internet, you can say google, but h exclude anything after twenty twenty three. Well, actually often give you Better results.

Yes, because is this much the explosion of AI generated materialism? crazy. So like in training rock, um we have to go through the data. And so I K we have have to have sort apply A I to the data to say, is this data most likely correct? Most likely not before he feared into the training system.

That's crazy. yeah. So you is a generated by humans. yeah. I mean that the data, the the data filtration process is extremely, extremely difficult.

Yeah, do thing is possible to have A A serious, subjective, rigorous political discussion with the rock. 呃， like for a long time. And he wouldn't like grab three or greg four.

Three is going to be next lever. I mean, what people are currently sing with rock is is kind .

of baby brock .

yeah baby it's baby brock, right? But baby are still pretty good um so it's but it's in order of money to less sophisti ted than GPT four now grog two, which finished training I don't know, six years go there are about ground will be a giant improvement and then grog three will be water my to .

Better than grog two and you hoping .

for IT to be like steady art, hopefully aspiration using a .

matters .

who mills asia the the people and how they think and how they structure the companies are all that kind of .

stuff yeah I think that matters that there is I think it's important that that the whatever A I went is a maximum is seeking A I that is not forced to live up local for me a reason really um political anything um. I I am concerned about A I succeeding that is that that has got that is programmed to lie, even if even is always.

right? Because in small ways becomes big way is .

when it's ways and when .

it's used more, more at scale by humans. Yeah since I am interviewing that little jump go.

you start by sure of stuff.

There was tragically an an assassination attempt on the trump um after this you twist that you endorse what's your philosopher behind the endorsement? What hope down from does for the future this country and for the future humanity?

well. I think this people can take like say, an endorsement as well. I agree with everything that persons ever done the entire life. One how to percent whole holy and that's that's not gonna true of anyone. Um but we have to pick you.

We got two choices really for for whose president and it's not not just who president but the entire mini structure uh changes over um and I thought truth displayed uh courage under fire object vely um you know he's a Scott shot Scott was dreaming down his face and he's like first pumping saying fine you know like that's impressive like you can't fain bravery in a situation like that um I must be would be ducking there would not be because I have a second child you know there were the president in our states can represent the country and um they're representing you they're representing a one amErica well thank you want someone who is strong and courageous h to represent the country um that's not to say that he is without flaws. We all have flaws but on baLance um and certainly the time IT was um a choice of, you know biden, poor, poor guy has trouble climbing flight test days once first pumping after getting shot no no comparison. I mean, who do you want dealing with um some of the tough as people, you know all the world leaders who are pretty tough himself.

And I mean, I say like what are the things that I think are important? Um you I think we want a secure border. We don't have a secure border.

We want safe and clean cities. I think we want to reduce the amount of spending that were at, slow down the the spending. And because currently spending at a rate that is background ting in the country, the interest payments on U.

S. Debt this year exceeded the entire defense department spending. If this continues, all of the federal government taxes will simply be paying the interest. And then, and you you can go down that road, you end up, you know, in the tragic situation that argentina have back in the day.

Argentino is to be on most prosperous places in the world and hopeful with malaya king over he can restore that but um IT was an incredible powful Grace for argentino to go from being one of the most prostate place in the world to um being very far from that. So I think we should not take american prosperity for granted. Um so we really want to I think we've got to reduce the size of government. We gotta reduce the spending and regarded within our .

means getting politicians in general, politicians.

governments.

how much power do you think they have to to steer humanity towards good?

Um I mean, there's a sort of agal debate in history like a you know is history determined by by these fundamental tides or is that determined by the captain of the ship? That's both really. I mean, there are tides and but also matters, his captain, the ship.

So through its post economy, essentially you does. But I mean, there are certainly tide. The tides of history are are, there are real types of history. And these these tides are often technologically driven. If you say, like the good bug press, you know, the widespread availability of books as a result of a pretty press, that that was a massive tide of history and dependent of any ruler. But you know, you, I install me times you want the best possible, capture the ship.

Well, for all, thank you for recommending a will in Ariel dances work. Have read the short one, but now the lessons of lessons of history is one of the one of the lessons, one of the things, the highlight is the importance of technology and technological innovation. And they, which is funny because they they wrote so long ago but they were noticing that the the rate of technological invitations speeding up um .

yeah I would .

love to see what they think about now um but yeah so he did to me the question is how much government, how much politicians get in the way of technological innovation building verses like help IT in which which, which politicians, which kind of policies help, help of connotation ation because there seems to be, if you look at a human history, that's an important component of empire rising and .

succeeding yeah.

Well, I an interactive .

dating civilization start civilization. I think the start of writing my views is the that's best by what I think is probably the the right starting point to date. Civilization and from the same point, civilization spent around for about fifty five hundred years when riding was invented by the ancient marions are gone now.

Well, but the the entrance area in most of getting a lot of fists, the those entrance mars really have a long list of fists. It's pretty well in fact, iran goes through the list. It's like you wannsee first. We will show you first the arians just as were just ascagne um and then the egyptians who are right next store um relatively speaking, like weren't that far, develop an entirely different form of writings. The hired gliff ics can inform and high clicks totally different.

And you can actually see the evolution of both high graphics and uniform, like the uniform thought of being very simple and then IT gets more complicated and then twenty and OK, they really get very investigated with the form. So I I think if the civilization is being about five thousand years old um and earth is um if this is correct, four half million years old, so civilization has been around four million about existence. Flash in the pan.

yeah. These are the early early days. And so we dry, we make IT very dramatic because has been rises and falls of empires, and many.

so many, so many rises in force of empire, so many.

And it'll be many more.

Yeah, exactly. I mean, only a tiny fraction proba less than one percent of of what would ever written in history is available to us. Now, mean, if they didn't put IT literally just late and stone, or put IT in a clay tablet, we don't habit.

I mean there's some small amount of like the paris goals that were recovered that a thousands of years old um because they were deep inside of paramount were affected by moisture uh but but but other that it's really gotta be in a plate outlet or chizzle. So the vast majority of stuff ff was not a chizzle because he takes a lot of the chisel things. Um so that's what we got tiny, tiny fraction of the information from history, but even that little information that we do have and the architecture record, uh shows so many civilizations rising and falling for sword.

We tend to think that were some hope different from those people. One of the other things that are highlights, as the human nature seems to be the same, just persist.

Yeah I mean, the basics of human nature are more less the same.

So we get ourselves in trouble in the same kinds of ways. I think even with the advances .

technology, yeah, I mean, you do tend to see the same patterns, similar patterns for civilizations where they go through a life cycle like like an organism, know of just like a human is sort of, I go with us, baby, you know, told ler, teenager, you know, eventually get get hold and dies. The civilizations go through a life cycle. No civilization will forever.

would you? What do you think he takes for the american empire to not collapse in the near term future, in the next hundred years, to continue flushing?

What the single list thing that .

is um but often .

actually not mentioned in history books, but that does mention IT uh is the birthright so um like like a path to some account in two of thing happens when civilizations become are are winning for too long that they they .

the birthrate declined.

I can often decline quite rapidly. We're saying that throughout the world today, there are currently south korea that I think maybe the lowest fatlings rate, but there are many others. They're close to IT as a point aid.

I think if the birthrate doesn't decline further, a south korea will lose roughly sixty percent of its population and and but every year that both way is dropping. And this is true through most world. And I don't mean single out south korea been happening through out the world.

So as soon as as soon as any given civilization reaches the level of prosperity, the birth rate drops. Um and you could look at the same thing happening in in each grown. So a really sees a technology this I think around fifty fifty H B C.

Um and try to pass one of your successful try to pass a law to give any center for any romance citizens that would have a third child. And I think Augustus was always able took what he was he at the dictators senate was just for show. I think you did pass a attacking centre of four roman citizen to have a third child, but those efforts we are unsuccessful. Rome fell because the roman stop having making romans. That's actually .

the fundamental issue.

And there were other things there .

was .

like they had like a quite a serious malaria, serious malaria epidemics, plagues and what not um for the head those before uh the the the the issues that the birthrate was follower than the .

death rate IT really is that simple that's more people acquired mental .

level if civilization .

does not at least maintain its numbers um IT will this sphere .

so perhaps the amount of computer that the biological computer allegations to sex is justified factors should probably increased.

Well, I mean, this heats tic sex, which is a the year that that's the hero there is.

It's not productive.

IT doesn't produce kid. Well, you what my what matters I mean, the rant makes this very clear, because he looked one civilization after, and they all went through the same cycle when the civilization was under stress. The birthrate was was high. But as soon as there were no extern enemies or they, they were, had a extended period of prosperity, the both rate inevitably dropped. Every time I put believe, with a single exception.

So that's like the foundation of IT. You need to have .

people yeah, I need a base level. Then no humans, no humanity.

And then there is all of things like, you know, human freedoms and just giving people the freedom to build stuff.

Yeah, yeah, absolutely this. But but at at, at a basic level, if you do not at least maintain your numbers, you should below replacement, right? And that trend continues, will eventually despair.

This is elementary. Now, then obvious also want to try to avoid like massive wars. You there's a global term nuclear war play were all toast radio actor toast. So we want to try avoid those things.

Um there are um there's a thing that happens over time with but then he given civilization, which is that the but law's regulations accumulate and if there is not if if there's not some forcing function like a war to clean up the accumulation of laws and regulations, eventually everything becomes legal and you the that's like the harden of the arteries um or a way to think of IT is like being tied down by a million little strings like gallagher if you can't move and I sound like any one of those strings is the issue which got many room so that has to be ache sort of garbage collection for laws and regulations so that you you don't keep accumulating laws and regulations to the point where you can do anything. This is why we can't build high speed rail in america. It's illegal. That's the issue. It's illegal six paces on sunday to build high rail in america.

I wish you could just like for a week going to washington and like be the head of the committee for making, uh, what is IT for the garbage collection, making government smaller, like removing stuff .

I have discussed with trap the idea of a government efficiency commission.

nice. yeah. And I would .

be willing to, uh, be part of that commission.

I wonder how hard that is.

The anybody reaction will .

be very strong.

So you you really have to you're tacking .

the matrix .

at that point. Matters will fight back.

How are you doing with that being attacked?

Me time? Yeah.

there's a lot of IT. Yeah.

there is a lot. I mean, every day I will set up, you know.

how do how do you .

keep your just positivity?

How do you optimism about the world, a clarity of thinking about the world. So just not become resentful or cynical or all that kind stuff just getting attack by you very much more people mr represented .

ah yeah that's that's a daily occurrence. yes. So um I mean IT doesn't me down at times. I mean makes me sad, but um.

I mean, at some point you have to sort of say, look the the attack of by people that actually don't know me um there and they try to genre clicks so if if if if you can of detach yourself somewhat emotion which is not easy um and say, okay look this is not yeah from someone that knows me or that they are they're literally just writing to get you know impressions and clicks um then you know then that I guess IT doesn't heard as much. It's like it's it's not quite water rower, duck back, maybe like as of duck back.

right? Well, that's good. Just about your own life, what you use a measure of success in your life.

Matter of success, I say, like what having useful things gna get done .

daily day basis, wake up in the morning. How can I be useful today?

Maximized utility area of the court of useful as very diff ult to be useful.

Let's scale as scale. Can you like speak to what IT takes to be useful for somebody like you? Well, there are so many amazing, great teams. Like, how do you locate your time to be the most useful?

Well, time? Time is the right time is the three currency. yeah.

So IT is tough to say what what is the best allocation time? I mean, there are in a often save should like looks at tesler tesla this year ll do over a hundred billion in revenue. So that's two billion dollars a week. If I make slightly Better decisions, I can affect the outcome by a billion dollars. So .

that 呢。

you know, try to do the best decisions I can not, on baLance, you know, least compared the competition, pretty good decisions. But the marginal value of of a Better decision can easily be, in the course of an hour, a hundred billion dollars.

Given that, how do you take risks? How do you do the the algorithm you mentioned? I mean, deleting, given a small thing, can be a billion dollars. How do you decide .

to well, I think I have to look at IT on a senate places because if you look at IT in the absolute terms, it's it's just I would never get to sleep but would just be like, I need just keep working and with my brain potter, you know and i'm not trying to get as much as possible out of this meat computer. So it's not it's pretty hard because you could just work all the time at any point like that. A slightly Better decision could be a hundred, a hundred million dollar impact for test, or spaces, for that matter, when IT is wild, when when considering the motile value of of time can be hundred hundred dollars now, or at times for more is .

your own happiness. Part of that equation of success .

IT has to be design degree, either from sad, from depressed, I make worse decisions. So I I can't have, like, if I have zero recreational time, then I make work worse decisions. So I don't have a lot, but it's of zero.

I mean, my motivation, if I got a religion of any kind, is religion of curiosity of trying to understand you, is really the the mission of the understand universe, trina, um or least set things in motion such that at some point civilization understands the universe or far Better than we do today. And even what questions to ask, as doulas Adams pointed out in this book, the some of the answers is ugly. The easy part to try frame the question correctly is the hard part. Once you frame the question correctly, the answer is often easy.

So um i'm trying .

to set things in motion such that we are, at least at some point people understand the universe so for space sets the goal is to make life multiple planetary .

um and which .

which which is if if you got these the firing paradox of where where the aliens, you've got these these sort of great fulcher like like why why have we not heard from the aliens? Not a lot. Lot of people think there are aliens among us. I often claimed every one. Nobody believes me.

but I did say .

alien registration at one point on my immigration document um yeah .

so I .

ve not seen any evidence billions so it's just that um this one one of the explanations is that a intelligent life .

is extremely .

rare um and again, you look at the history of earth, civilizations only been around for one million, the first existence so if if he had posited here hundred thousand years ago who like well that I even have writing you know just how other basically so so.

How long does this civilization last? So for SpaceX the goals to stablish a self sustaining city on mars mars only viable planet for such a thing um the moon is close but its a less resources and I think is probably vulnerable to any any any calamity that takes out earth the moon is too close and it's vulnerable to get A A city to takes out of so nothing we couldn't have a moon base but mars is mars reform a resilient um the difficulty of getting to mars is what makes IT resilient so but you know in in going through the diverse explanations of why don't we see the aliens why one of them is that they they fail to pass these. These great filters, these these key hurdles, and one of those hurdles is being a multi ets species.

Um so if you're mulvaney's es, then if something would happen, whether that was a natural catastrophe or AManda catastrophe, he at least the other planet would probably still be around. So you're not like you don't have all the x one basket. And once you are sort of a two plant species, you can obvious ly extend to life, has to the asteroid belt to maybe to the moons of jupiter and ultimately need to other star systems. But if you can even get to another planet to the.

you know .

definite not getting .

to star systems, any other possible great filters, super powerful technology like agi example. So you you're basically trying to not go one great filter the time.

Digital super intelligence is but possibly a great filter. I hope IT isn't, but IT might be. You know, guys like, say, jeff hinton would say, you know, has he invented another? The key principles and artificial intelligence I keys puts the probability of A I annalist around ten to twenty percent, someone like that. So you know.

so it's it's .

not to sound like you look at the right side. It's ABS and likely to be great. So so but I I think I rist medication is important. Being a multipass species would be a massive risk mitigation and um I I don't anna serve once getting ever size is important the importance of having enough children to sustain our numbers are not going not plumb into population collapse, which is currently happening.

Population collapse is a real and current thing um so the only reason is not being reflected in the total population numbers is that is that as much is because people living longer um but but you you see to predict so what the population of any given country will be. Um you should take the birthrate last year, how many waves were born. I'll apply that by life experienced and that's what the population will be a study state unless if if the birthrate continues at that level. But if IT keeps declining, IT will be even less and eventually bunt to nothing so I keep you know banging on the baby drum here um for a reason um because IT has been the the source of civilization collapse over over again a throughout history. And so one way just not try try to stable for that day.

Well in that way I have miserably failed civilization and I am trying hoping to fix that. I'd love to have many kids great.

Hope you do um that time like the present yeah now .

I got to allocate more compute to the whole process um but apparently .

is not that difficult no it's like .

uncle labor um well, if I one of the things are you do for me, for the world is to inspires with what the future could be. And so some of the things we've talked about, some of the things you're building, alleviating human suffering with neural link and expanding the capabilities of human mind, trying to build the colony on mars, so creating a back up for humanity on another planet, and explain the possibilities of what artificial intelligence could be in this world, especially in the real world. Ai, with the hundreds of millions, maybe billions of robots walking around.

there will be billions robots. That's a seems a certainty.

Thank you for building the future. Thank you for inspiring so many of us to keep building and credit cool stuff, including kids. They go for .

a multiple.

go forth, multiple. Thank you. You want thanks to talk about them.

Thanks for listening to this conversation with elan mosque. And now, dear friends, here's D. J. Saw the co founder, president and C, O, O of new link. When did you first become fascinated by the human brain?

For me, I was always interested in understanding the purpose of things and how IT was engineer to serve that purpose, whether it's organic or inorganic. You, like we were talking earlier about your curtain holders, they serve a clear purpose, and they were engineered with that purpose in mind.

And in a growing up, I had a lot of interest in seeing things, touching things, feeling things and trying to really understand the root of how IT was designed to serve that purpose. And you know, obviously, brain is just a fascinating organ that we all Carry. It's a infinitely powerful machine that has intelligence and cognition that arise from IT.

And, you know, we haven't even scratched the surface in terms of how all that occurs. But also at the same time, I I think IT took me a while to make that connection to really studying and building tech, to understand the brain. But not until graduate school.

You know, there are couple of moments, key moments in my life where some of those, I think, influence how the trajectory or in my life got me to studying, uh, what i'm doing right now. You know, one was growing up both sides of my family. Uh, my grandparents had a very severe form of alzheimer a and it's incredibly debilitating ing conditions. Um I mean, literally you're seeing someone's whole identity and and their mind just losing over time. And I I just membered thinking um how both the power of the mind, but also how something like that could really boost your sense of identity is fascinating.

That that is one of the ways to reveal the power thing by watching IT lose the power.

And a lot of what we know about the brain actually comes from, uh, these cases where there are trauma to the brain, or some parts of the brain that let someone to lose certain abilities and as a result, there are some correlation and understanding of that part of the tissue being critical for that function. And um it's an incredibly fragile organ, if you think about IT that way, but also it's incredibly plastic and incredibly resilient in many different ways.

And by the way, the turn plastic will use a bunch means that it's adaptable. So new plastic refers the adapt ly the human brain, correct?

Um another key moment that sort of influence how the trajectory of my life have shaped towards the current focus of my life has been during my teenager when I came to the U S.

And I didn't speak award of english there was a hugely english barrier and um there is a lot of struggle to to connect with my peers around me um because I didn't understand the the artificial construct that we have created a called language, a specifically english in this case and a member of feeling pretty not being able to connect with peers around me. So spent a lot of time just on my own in reading books, watching movies um and I I naturally sort of gravitates towards size I books. I just found them really, really interesting and also was a great way for me to learn english.

Some of the first set of books that I picked up our enders game, you know the whole saa by uh worsen Scott card and neuromancer from Molly gibson and snow crash from neil Stevenson and you know, movies like matrix was coming out around that time point that really influence how I think about the potential impact the technology can have for our lives in general. So fast track to my collection. No, I was always fascinated by just just physical stuff, building physical stuff and especially um disco things that had some sort of intelligence.

And you know I started ed eleven during during undergrad and I started out my research in mens so microelectrodes accountable systems um and really building these tiny nano structures for a temperature sensing. And I just found that to be just incredibly rewarding and fascinating subject to just understand how you can build some something minature like that that again serve a function and head purpose. And then you know I I spent large majority of my college, yours basically building no limit orr wave circuit for next general communication system for imaging.

And IT was just something that I found very, very intellectual, interesting. You phase or raise how the the signal processing works for any modern, as well as next year telemundo ation system, wireless and wireline em waves or electro magnetic waves are fascinating. How do you design antennas that are um most sufficient in a small for print? How do you make these things energy efficient? That was something that just consume my intellectual curiosity.

And that journey let me to actually apply to and find myself in P H D. Program at u berkeley, at kind of this consortium called the berkeley wireless resort center that was precisely looking at on building at the time we call that X G, you know, similar to 4G， 4G5g by the next next generation g system and how you would design circuits around that to ultimately gone phones and you know basically any any other devices that are wireless to connected these days。 Um so I I was just absolutely just destination by how that entire system work and that in infrastructure works.

Um and then also during grass co, I had sort of the fortune of having um you know couple of ref fellowships that let me to pursue whatever project that I want. And that's one of the things that I really enjoyed about my graduate school career where you got to kind of pursue your intellective curiosity and the domain that may not matter at the end of the day but is something that really uh allows you the opportunity to um go as deeply as you want as well as as widely as you want. And at the time, I was actually working on this project called the smart banded.

And the idea was that and you get a wound, there's a lot of other kind of liberation of signaling pathway that cells follow to close up point. And there were hypothesis that when you apply external electric field, you can actually accelerate the closing of that field by having basically electoral taxing of the cells around that woodside. And specifically, not just for Normal moon, there are chronic moons that don't heal. Um so we were interested in building in a some some of a variable patch that you could apply to kind of facilitate that healing process. And that was in collaboration with uh professor mishaal maher's its um you which which you know was a great addition to kind of my teeth s committee and you know really shaped rest of my peach to career so this would be .

the first time you interacted with biology, I suppose.

Correct correct. I mean, there were some preferable, you know, in the application of the wireless imaging until communication system that I was using for security and bio imaging. But this was a very clear direct application to bio biological and biological system and understanding the contracts around that and really designing and engineering electrical solutions around that.

So that was my first introduction and that's also um kind of how I got introduced to Michelle. Um you know he sort of known for remote control of beetles in the early two thousands and then around twenty thirteen you know obviously kind of the holy grail when IT comes to implantable system is to kind of understand how small able thing you can make. And a lot of that is driven by how much energy or how much power you can supply to and and how you extract data from IT.

So at the time at berkely, there was kind of this uh, desire to kind of understand in the neural space what what what sort of system you can build to really militarize these implanted able systems. And a distinct tilly member, this one, particularly meeting or mishal, came in and his like, guys, I think I have a solution. The solution is of and uh and then he proceeded to kind of walk through why that is the case.

And that really formed the basis for my this is work um uh called neural dust system that was looking at ways to use ultra sound as opposed to uh electro magnetic waves for powering as well as communication. I guess I just step back and say the the initial. Goal of the project was to build this tiny, about the size of a neuron implantable system that can be part next to a new, on being able to record its date and being able to pink that back to the outside world, doing something useful. As I mentioned, the size of the implantable system is limited by how you power the thing and get the data of that.

And at the end, they fundamentally, if you look at a human body, where are essentially bag of salt water with some interesting proteins and chemicals, but is mostly salt water that's very, very well temperature regulated at thirty seven, two weeks years um and we'll get into how, why and and later why that's an extremely harsh environment ment for any electronics to survive as i'm sure you've experience or maybe not experiences in a dropping cell phone in in a salt water in an ocean IT will instantly cut the device right um but anyways but just in general, electro magnetic grapes don't penetrate through this environment well. And just the speed of light IT is what IT is. We can we can change IT.

And based on the the wavelength which you are interfacing with the device, the device needs to be big, like these inductor need to be quite big. And the general good rule of thum is that you want the way front to be roughly on the order of the size of the thing that you're interfacing wit. So an implantable able system, uh, that is around ten, two hundred micron in dimension, in in, in a volume, which is about the size of the neuron that you see in in a human body. And you would have to Operate that like hundreds of giga hurts, which number one, not only is that difficult to build electronics Operating at those frequencies, but also the body just a ten ways start very, very significantly. So the interesting kind of insight of this ultrasound um was the fact that ultra sound just travels a lot more effectively in the human body tissue compared to electro magnetic waves.

And this is something that you encounter uh and you i'm sure most people have encounter in their lives when you go to um in a hospital, there are medical uh watch sound in a snowpack right um and they go into very, very deep death without attending rating much too much of the signal so all in all, in a vulture, a sound, the fact that IT travels through the body extremely well and the mechanism to which IT travels to the body really well is that just the way front is very different. It's uh electromagnetics waves are transfers, whether an ultrasound waves are compressive. So it's just a completely different mode of uh wave front propagation um and as well as speed of sound is orders and orders of magnitude stand speed of flight which means that even at ten mega hearts watter sound wave, your wave front ultimately is a very, very small wavelength th so if you are talking about interfacing with the ten micron or hundred micron type structure, you would have hundred sixty micron away from a ten meghan ts.

And building electronics at those uh at those frequencies are much, much easier and there a lot more efficient. So the basic idea kind of was born out of um you know using ultrasound as a mechanism for pouring the device and then also getting data back. So now the question is, how do you get the data back? The mechanism to which we landed on is what's called back sketcher's.

Um this is actually something that is very common and that we interface on the data basis with our R F I D cards, you know, a radio frequency I D tax where there's actually rarely you know in your I D A battery inside there's an antenna and there's some sort of uh coil that has your serial uh identification idea and then there's an external device called the reader that then sends a way front and then you reflect back that way front with some sort of modulation is unique to your I D A that's what's called back sketcher's uh, fundamentally. So the tag itself actually doesn't have to consume that much energy. And um that was a mechanism till which we were kind of thinking about sending the data backs when you have an external uh ultra sonic transducer that sending ultra sonic wave to your implant. The neural dust in plant and IT records some information about its environment, whether the new on firing or uh some other state of um the uh the tissue that is interfacing with and then it's just emptied module the way front that comes back to the source and the recording .

step would be the only one that requires an energy. So what would require energy that step?

correct? So IT is that initial kind of start up circuitry to get that recording and providing IT. And then just module and the mechanism to which that that you can enable that is there is the specialized Crystal called P S electric Crystals that are able to convert sound energy into electrical energy and vice for so you can of have this inter play between the volta onic c domain and electrical domain, that is the the biological tissue.

So on the theme of parking, very small computational devices next to unions. That's the dream of the vision of brain computer interfaces. Maybe before we talk about your link, you give a sense of the history of the field of B C. I, what war has been um maybe the continued dream and also some of the most stones along the way of the different approaches and amazing work down at the various labs.

I think a good starting point is um going back to seventy nineties.

do not expect that where .

the concept of animal electricity or the fact that body's electric was first discovered by louie gobi where uh he had this famous experiment where he connected side of electrodes to frog leg and rent current road and then I started twitching and he said, oh my godness boy's electric ah so fast for many, many years to nineteen twenty years uh where hands burger, who is a german psychiatrist, discover E E G or electoral h in pilot phy, which is still around there. These are electoral race that you wear outside the score that gives you some sort of neural recording that was a very, very big milestone that you you you can record some sort of activities about the human mind. And then in the ninety eighties there were uh this group of scientists, renta forms and morson that um inserted these glass .

micro electrodes .

into the cortex and record a single new the fact that they they there's signal that are uh a bit more high resolution and high quality uh as you get causer to the source, let's say. And in the nineteen fifties um these two scientists hutchinson in hawksley shot up and they are built this beautiful beautiful models of the cell membrane and the ironic mechanism and had this like circuit dirac and as someone who is an engineer, it's a beautiful model that in a built out of these partial differential equations talking about flow of ions and how that really leads to how neurons communicate and they want the nobel Price for that ten years later in the eighteen sixties so in nineteen sixty nine uh efforts from university of washington published this beautiful paper called Operant conditioning of critical n activity where he was able to record a single unit neuron from a monkey and was able to have the monkey modulated based on its activity and reward system. So I would say this is a very, very first example on, as far as i'm aware, of close loop ua brain, computer face or bci.

The abstract reads the activity of single neurons in personal cortex of and exercise monkeys was conditioned by reinforcing high rates of neuronal discharge with delivery of a food pot. Auditory and visual feedback of unit firing rates was usually provided addition to food reinforcement. cool. So they actually gotten done.

They had done. This is a back in one thousand nine and sixty nine.

after several training sessions, monkeys could increase activity of newly isolated cells by fifty to five hundred percent above rates before reinforcement.

Fascinating brain is very plastic and so .

and from here the number of experiments grew yeah number .

of experiments as well such a tools Turner face with the brain have just exploded. Um I think and and also just understanding the neural code and how some of the cortical layers and and the functions are organized. So the other paper that is um uh a pretty seminal, especially in the the motor decoding uh was this paper in the eight thousand nine and eighties from georgia Operas um that discovered that there is distinct LED motor tuning curve.

So what are motor tuning curve is the fact that there are in the neurons, in the motor cortex of numbers, including humans, that have a preferential direction that causes them to fire. So what that means is there are set of neurons that would, uh, increase their spiking activities. When you are thinking about moving to the left, right, up, down, and any of those are vectors. And based on that, you know, you can start to think, well, if if you can identify those essential ign doctors, you can do a lot, and you can actually use that information for actually decoding someone's intended movement from the cortex. So that was a very, very seminal kind of paper that showed um that uh there there is some sort of code that you can you can attract, especially the motor cortex.

So there's signal there. And if you measure a the the electrical signal from the brain that you can, you can actually figure out what the intention .

was cry yet not only electrical signals, but electrical signals from the right side of neurons that give you this preferential direction.

okay. So going slowly towards neurogen. link. One interesting question is what you understand on the B C, I front on invasive versus not invasive from this line of work? Uh, how important is IT to to park next to the neuron? What does that get you?

That answer fundamentally depends on what you want to do with that, right? Um there's actually incredible amount of stuff that you can do with E E G N A electric cordial graph, e coc, which actually doesn't penetrate the the cortege laure prankt, but you place a set of electrics on the surface of the rain. So the thing that I personally very interested in is just actually understanding um and in being to just really tap into the high resolution, high fidelity understanding of the activities that are happening at the local global.

And you know we can get into bio physics, but just to kind of step back um to to use analogy because analogy here can be useful sometimes it's a little bit far ult to think about electricity um at the end day we're doing electrical recording that mediated by iron um current in the movement of these charge particles um which is really, really hard for most people think about but turns out a lot of the activities that are happening in the brain and the frequency band with which that happening is actually very, very similar to sound waves and and in our Normal conversation an audible range range. So the analogy that typically is used in the field is. And if you if you have a football stadium, uh, you know there's game going on.

If you stand outside the stadium, you you maybe get a sense of how the game is going based on the cheers and the boost of the home crowd, whether the team is winning or not. But you have absolutely no idea what the score is. You absolutely no idea um individual audience or the players are talking or saying to each other what the next players, what the next school is um so what you have to do is you have to drop the microphone near into the stadium and then get near the source like to the individual chatter um in this specific example you would want to have IT you know right next to where the hoddle is happening um so I I think that's kind of a good illustration of what we're trying to do. Um when we say invasive or minimally invasive or implant brain, computer interfaces for is non invasive or non implant brain in their faces. It's basically talking about where do you put that microphone and what can you do with that information.

So what what is the bio physics of the read right communication that we're talking about here as we now step into the efforts of neural link?

Yeah, so uh, brain is made up of these specialize cells called news. There's billions of them, you know, tens of billions, you know sometimes people called hundred billions that are connected in this complex yet dynamic network. Uh they're constantly remodeling. You know they are changing their senates ites um and that you know will we typically called neuroplasticity.

And the neurons are also bathe in this charged environment that is a latent with many charged molecules like potassium ions, sodium ions, choring ions and uh those actually facilitate this, uh you know through I on a current the communication between these different networks. And uh when you look at the look at a neuron as well, um they they have these are membrane, a beautiful beautiful uh protein structure called the vote ge selective ion channels, which in my opinion is one of natures best inventions in many ways. If you think about what they are, they're doing the job of a modern day transistors.

Transistors are nothing more at the end of the day than a voltage gated conduction channel um and nature found a way to have that there very earlier on in the evolution. And as we all know, with the transfer, you can have many, many competition and a lot of amazing things um that that we have access to today. So I I I think it's one of those just as a attention, just beautiful, beautiful invention that the nature came up with footage of on channels.

I mean, I suppose those are on the biological of the every level of the complexity of the higher chy of the organism. There's going to be some mechanisms for storing information and for a doing computation, and this is just one such way. But to do that with the biological and chemical components is ensuring plus like when neurons, I mean, it's not just electricity, uh, chemical communication service, mechanical and these are like actual objects that have like that vibrate, I mean they move. Yes, there are actually .

I think there's a lot of really, really interesting physics that that that are involved. And you know kind of going back to my um work on ultra sound h in grass school, there there are groups and uh there were groups and there are still groups are looking at ways to causing irons to actually fire and action potential using little on wave and the mechanism to which that happening is still unclear as I understand. Um you know you may just be that you know you're imparting some sort of thermal energy and that causes cells to depolarizer some interesting ways um but there are also these um iron channels or even membranes that actually just open up is poor as there are being mechanically like shock right vibrated so there's just a lot of elements of these like move particles um which again like that govern by division physics, right movements of particles and there's also a lot of kind of interesting physics there also .

not to mention as Roger peno stocks about the there might be some beautiful warehouse in the quana mechanical effects this and he actually believes that conscious this may emerge from the quality mechanical effects there. So like there's physics, there's chemistry, there's bio, all of that is going .

on there oh yeah yeah I mean, you can yes, there's there's levels of physics that you can dive in to, but yeah in the end you have these um uh members with these voltage gated eye channels that selectively LED um discharge molecules that are in in the extra solution matrix like in and out.

And these neurons generally have these like resting potential where there's a voltage difference between side the cell and outside the cell and when there's some sort of stimuli that changes uh the state such that they need to send information to the the downstream network. Um you know we started to to see these like sort orchestration of these different molecules going in and out of these channels. They also open up like more of them open up wants to reach some threshold uh to a point where you know you have a depolarizer cell.

的 选择 真 不太 熟， so it's a just a very beautiful kind of orchestration of these are these these molecules. And um what we're trying to do when we place an electorate or parking IT next to a neuron is that you're trying to measure these local changes in the potential um again mediated by uh the the um the movements of the islands. And what's interesting as as a mention early, there's a lot of physics um and the two dominant physics for this electrical recording domain is diffusion physics and electro magnetism.

And where one dominate, where my maxwell S A equation dominate versus fix law dominate depends on where your electrode is um if it's close to the source, a mostly electromagnet based. Um when you're further away from IT, it's more division based. So essentially .

when you're .

able to park IT next to IT, you can listen in on those individual chatter and those local changes in the potential and the type of signal that you get are this canonical textbook neural uh spiking way from when you the moment you're further away and based on some of the studies that people have done um in a Christopher coke lab and and others， once you away from that source by roughly around hundred micron, which is about with about human hair, you no longer here from that neuron.

You are no longer able to kind of have the system sensitive enough to be able to a record that particular um local membership potential change in that neuron. And just to kind of give you a sense of scale also when you when you look at a hundred micron box or hundred microns by hundred microns by hundred micron box uh in a brain tissue, there's roughly around forty new on and whatever number of connections that they have. So there's a lot in that volume of tissue. So the moment you are outside of that, you there's just no hope that you'll be able to do detect that change from that one specific around the you make .

care about yeah, but as you're moving about the space, you'll be hearing other ones. So if you move another hundred, Michael, you'll be hearing chat for another command, correct? And so the whole senses, you want to place as many as possible electrodes. Then you listen to the chatter.

Yeah, you want to listen to the chatter and and the end of the day, so want to basically let the software do the do the job of decoding and just a kind of go to in why E O G E G work at all, right? Um when you have these local changes, you know obviously it's not just this one you're on that's uh activating. There's many, many other networkers are activating all the time.

And you do see sort of a general change in the potential of this elector like this is charge medium and that's what you are recording when you're farther away. I mean you still have some reference elector that's a stable and the brain that just electro active organ and you're seeing some combination aggregate uh action potential changes and then you can pick IT up right is a much slower um changing uh signals. But you know, uh there there are these economic kind of isolation and waves like gma waves, bet ways is like when you sleep that can be detected, the sort of a sync ony ze um kind of global global uh effect of the brain that that you can detect.

And I mean the physics of this go like I mean if we really want to go down the rabbit hole like there, there's a lot that goes on in terms of like why diffusive physics at some point dominate when you are further away from the source. You know it's just a charge medium um so similar to how when you have electro magnetic ways propagating in atmosphere or in in a charge medium like a plasma, there's this wear shield thing that happens that actually um further attention to signal um as you move away from IT. So yeah you see, like if you do a really, really deep dive on kind of the signal tannian over distance, you start to see kind of one of our r square in the beginning and the exponential drop off. And that's the knee at which, you know, you go from electric magnetic m dominating to diffusion physics dominating .

but once again with the electrons the by physically you to understand this is um not as deep because no matter what you're placing that you're listening to a small .

crowd of local neurons, correct? So when you penetrate the brain big and .

there's a lot of neurons, there are any many of them. But then again, there's like there's a whole feel the new science is studying like how the different groupings, the different sections of the seeding in the rena, what they usually responsible for, which is where the the metaphor probably falls apart. The thing is not that also .

most of them are silent. They not really do much you know or or they their activities or um you know you have to hit with just the right side of stimulus.

So they're usually quiet.

They're usually very quiet, quiet. There's I mean, similar to dark energy and dark matter, there's dark neurons. What are they all doing when you place these elector again?

Like within this hundred micon volume, you have forty or so neurons. Like what do you not see? forty? What do you see? Only a handful.

What is happening there? Well, they're most acquired. But like when they speak this, a profound shit, I think as though i'd like to think about IT anyway before we zoom in even more as zoom out.

So how does neural link work from the surgery to the implant to the signal and the decoding process and the human being able to use the airline to actually affect the the world outside? And all of this are asking in the context of there's a gigantic historic mouths on the new link. Just accomplished the january of this year, uh, putting in your link implant in the first human being, no land. And it's been a lot to talk about there about his experience because he is able to describe all the new nuances in the facing complexity of that experience of everything involved. But on the technical level, how does neural link work?

Yes, there are three major components to the technology that we're building. A one is the device um the thing that's actually recording these neural chatters, uh we call IT in one in plant or the link and uh we have a surgical robot that's actually doing an implantation of these tiny, tiny wires that we call threats that are in a small than uh human hair and once everything socialized, you have these neural signals, this spiking neurons that are coming out of the brain.

And uh you need to have some sort software to decode what the users intend to do with that. Um so there's what called the nearing application or be one APP that's doing that. Translation is running the very, very simple machine learning model that decodes these um inputs that are neural signals and then converted to set of outputs that allows um in our participant first participant knowen to be able to control a curse.

And this is done wirelessly and .

this is done wirelessly. So we our our implant is actually two part that this link has a uh you know these flexible tiny wires core thread um that have uh multiple electrodes along its links and um there are only inserted into the cortical layer which is about three to five millimeters in a human human brain um in the motor cortex. Reason that's where the kind of the intention for movement wison and we have sixty four of these threats each thread having sixteen electrodes along you know the span of three to four millimeters um separated by two hundred and micron so you can actually be cord along the death of the insertion. And based on that signal uh there's custom um you know um innovate circuit or a sector we built that emphasizes the neural signals that are recording and then dividing and then um has some mechanism for detecting whether there is a an interesting event that is a spiking event um and decide to send that or not, send that through blue tooth to an external device, whether it's A A phone or a computer that's running this only application.

So is on board signal processing already, just decide whether this is an so there is some competition power on board inside the in addition to the human brain.

Yeah so IT does the signal processing to really impress the amount of signal recordings we have a total of thousand electrodes um sampling at uh you know just under twenty kilo hurts with temped each. So uh that's two hundred make a bit that's coming through to the trip uh from thousand uh channel. Some is uh neutral recording and that's quite a bit data.

And you know there there are technology available to send that off wirelessly. But being able to do that, a very, very firmly constraining environment that is a brain. So there has to be some amount of compression that happens to send off only the interesting data that you need, which in in this particularly case for motor ity coding is um occurrence of a Spike or not. And then um being able to use that to a to a you know decode the intended cursor movement so the implant self processes IT figures out whether Spike happened or not with our speke detection algorithm and then sends out off packages IT sends off through blue tooth um to an external l device that then has the model to decode. Okay, based on the spiking inputs, the knowen wish to go up, down, left, right or click or right click or whatever.

all this is fascine, but let's stick on the end when he plant itself. So the thing that's in the brain ah i'm looking at picture of IT, there's an enclosure, uh, there's a charging cause. So we didn't talk about the charging, which is fascinating.

Uh, the battery, the power electronic, the anti na, uh then there's the signal process electronics. I wonder if there's more kinds of signal process you can do IT. That's another another question.

And then there's the threads themselves with the enclosure on the bottom. So maybe to ask about the charging. Yeah, there's an external charging device.

Yeah, there's an external charging device. So the second part of the implant, the threats are the ones again, just the the last three to five millimeters are the ones that are actually penetrating the cortex.

Uh rest of IT is actually most of the volume is occupied by the battery uh rechargeable battery um and uh you know about the size of a quarter, you know actually have a device here if you want to take a look at IT, no, this is the the flexible threat component of and this is the implant. So it's about the size of A U S. Quarter um about nine millimeter thick.

So basically this implant, uh you know want you have the crane action and the and the directory um ths are inserted and um the the whole that you create, the screening academy gets replaced with that. So basically that clock that hole and you can screw in uh these self drilling cranial screws to hold them in place. And at the end of day, once you have the skin flap over a there's only about two, three millimeters that's you know obviously transitioning off of the top of the implant to where the screws are and and that's the minor bump that you have.

Those threads look tiny. That's incredible. That is really incredible. That is really incredible. And also, as you're right, most of the value actual volume is the battery. Yeah, this is waste more than I realized.

They are also the threads themselves .

are quite strong. They look strong.

And in the thread themselves also has a very interesting um feature at the end of IT called the loop, and that's the mechanism to which the robot is able to interface. And manipulative tiny airline structure and their .

time is so what's the wife of a thread?

Yeah so the the wife of a thread um starts from sixteen micron and then tapers out to about eighty four micron. So you know every human hair is about eighty two hundred my grand with.

This thing is amazing. This thing is amazing.

Yes, most of the volume is occupied by the, by the barry we charged with my own. So and a the charging is done through inductive charging， which is actually very commonly use, you know, your cl phone, most cl phones have that. The biggest difference is that you know, for us, usually when you have a phone and you want to to charge IT on the charging pad, you don't really care how hot IT gets where.

As for us, IT matters, there is a very strict regulation and good reasons to not actually increase the surrounding tissue temperature by two. So there's actually a lot of innovation that is packed into this to allow charging of this implant without causing that temperature thresh to reach. And even small things like you see this charging coil and what's called a favorite shield, right?

So uh without that fair a shield what you end up having when you have um you know resent inductive charging is that the battery yourself is a metallic in and you form these eight current um from external charger and that causes heating um and that actually contributes to inefficiently and charging um so this very right. sure. What he does is that is actually concentrate that fields line away from the battery and then around the coil that actually wrapped around IT.

There's a lot of really faster design here to make IT. I mean, you're integrating a computer into a biological a complex biological system. Yeah.

there's a lot of innovation here. I would say that part of what enabled this was just the innovations in the wearable. Uh, there's a lot of really, really powerful tiny low power uh, micro controllers, temperature sensors or various different sensors and power electronics. A lot of innovation really came in the the charging coil design, how this is package and how do you enable charging such that you don't really uh exceed that temperature limit, which is not a constraining for other devices out there.

So talk about the threats themselves, those tiny, tiny, tiny things. So h, how many of them are there? You mentioned a thousand electoral. How many threads are there? And what did the electrons have to do with the thread?

yeah. So the current instantiation of the device has sixty four threads and each thread has sixteen electrodes for a total of one thousand twenty four electrodes that are capable of both recording and stimulating. And the thread is basically this polymer insulated wire um the metal conductor is the kind of a chemise cake of a thai pat goal pat thi um um and they're very, very tiny wires on two micron in with so two one million of uh meter. It's crazy that thing .

i'm looking at IT has the party er insulation, has the conducting material and has sixteen electrics at the end of IT. On each of those thread yeah on each of those threads.

correct sixteen each one of us.

you and that I mean, to state the obvious or maybe for people who are just listening.

they're flexible. yes. Yes, that's also one element that uh was incredibly important for us. So each of these thread are I mentioned sixteen micron and wife and then they take her to eighty four micron. But in thickness there are less than five micron um and in thickness is mostly you know a polyamine at the bottom and this metal track and then another polymers, so two micron of polymers, four hundred anoma ter of this metal stack and two micron of polyimide sand which together to protect them from the environment that is uh thirty seven degrees see bag of salt water .

so with some maybe can you speak to some interesting aspects of the material design here? Like what does IT take to to design a thing like this, to be able to manufacturer thing like this for people who don't know anything about this kind of thing?

yeah. So the material selection that we have is not I don't think IT was particularly unique. Um there there were other labs and there are other labs that are kind of looking at similar um material stack. Um there's kind of a fundamental question um and and still needs to be answered around the longevity and reliability of these uh micro lectures um that that we call compared to some of the other more conventional neural interfaces devices that are intra kenion so penetrate cortex that are more rigid um you know like the u tory um that that are these uh four by four millimeter kind of silicon shank that have exposed uh recording side at the end of IT um and and um you know that's that's been kind of the innovation from Richard Norman back in one thousand and ninety seven. Uh it's called the uta because you know he .

was a university of utah and what what is the utah rate look like as a rigid type of.

yes, we can actually work IT up.

Yeah, yeah.

it's a Better needle. Um there's. Okay, good. Those are gig. Yeah you want. And the size and the number of shanks vary anywhere from sixty four to one twenty eight um at the very tip of IT is an exposed electoral that actually records all signal.

Um the other thing that's interesting to note is that a unlike neurolinguistic s that have recording electrodes that are actually exposed radium oxide recording sites along the death, this is only at a single death. So these utility spokesman be anywhere tween point five millimetre to one point five millimetre. And they also have uh designs that are slanted um so you can have IT inserted at different death um but that's one of the other big differences. And then uh I mean the main key differences, the fact that uh there's no active electronics, these are just electrodes and then there's a bundle of a wire that you're seeing and then that actually then exits the cranny academy um that then has the support that you can connect to um for any external electronic devices they are working on or have the wireless kilometres device. But I still requires us through the skin uh port that actually is one of the biggest failure remotes for induction uh for the system.

What are some of the chAllenges associate with flexible threads? Like for example, on the robotic side are one uh in planting those threads how difficult discarded task .

yeah um so as you mention they are they're very very the new where by hand on this uta that you you saw uh earlier, they are actually inserted by a newer surgeon actually positioning IT near the site that they want and then uh there actually there is a new matic camera that actually pushes them in.

Um so so it's uh it's a pretty a simple process um and they are easier maneuver um but for this inform a raised, they are very very tiny and a flexible so they they are very difficult. That's why we built tough robot to do that. Um there are other other reasons for why we built the robot um and that is ultimately we want this to help millions and millions of people that can benefit from this.

And there are just aren't that many new resources out there. And uh you know robots can be uh something that you know we hope can actually do large parts of the surgery. Um but yeah ah the robot is the entire other um sort of category of product that we're working on. And IT is essentially this mult acis gentry system that has the specialized robot head um that has all of the optics and um this this kind of a needle retracting mechanism that menu these threats um via this loop structure that you have on the thread. So the thread already has a .

loop structure about which you can grab IT correct correct. So this is that thing. So you mention optic, so there's a robot r one.

So for now, there's a human that actually quotes A A hole in the in scope. And then after that, there is a computer vision component that's finding a way to avoid the blood vessels. And then you're grabbing by the loopy individual thread and placing IT in a particular and to avoid the blood vessels. And also choose in the depth of placement, all the process of controlling every, like the three d geometry of the placement, correct?

So the the aspect of this robot that is unique is that it's not surgeon assisted or human assisted. It's a my automatic automatic uh robot once you you know obvious ly, there are human component to IT. When you're placing targets um you can always move IT away from kind of major that s that you see um but I mean, if we wanted get to a point where one click and IT just does the surgery within minutes.

So the computer vision component finds great targets, candidates and the human kind of approved them and robot is does do like one at a time.

does do IT one at the time. Ah and that's actually also one thing that we uh are looking at, ways to do multiple threats at a time. There's nothing stopping from IT. You can have multiple kind of engagement uh mechanisms um but right now it's one by one. And the you know we also still do quite a bit of just kind verification to make sure that I got inserted if so, how deep actually match um what was programme then and so .

and so th and the the actual lecture a place the very at different depth in the I couldn't I mean is very small differences .

but differences .

yeah yeah and so that there's some reasoning behind that as you mention like IT IT gets more very signal yeah .

we I mean we try to place them all around three or four millimeter from the surface um just because the span of the electoral, those sixteen electors that we currently have in this version spans um you know roughly around three millimeters. So we want to get all of those in the brain.

This is fascine. okay. So there's a million in questions here that if a goes dome is specifically the electoral, so what is your sense how many neurons is each individual elector .

listening to? Yeah each electoral can record from anywhere between zero to forty, as I mentioned right earlier. Um but practically speaking uh we only see about and most like tutor um and you can actually distinguish which neuron is coming from by the shape of the Spikes. So I mention the Spike detection algorithm we have is called boss algorithm. But for online Spike sorter nice IT actually outputs at the end of the day uh six unique values which are um you know kind of the impetus of this like negative going hump， middle hump like uh positive going hump. And then also the time at which this happen and from that you can have of a statistical probability um estimation of is out of Spike is and out of Spike and then based on that you could also determine oh, that Spike looks different than that bike must .

come from a different year on o that's a nice signal process step from which you can then make much Better predictions about if there is, especially in this kind of context, there could be multiple new screaming and that that also results in you being able to compare the date to Better.

Yeah sunday okay that and just to be clear, I think the loves do this what's called Spike sorting um usually once you have this like broadband, you know like the fully digitize signals and then you run, I bunch a different set of algorithms to kind of teeth apart. Is just all of this for us is done on the device, on the dev, you know very low power custom you know built asic digital processing unit.

Highly he constrained.

highly he constrained. And the processing time from signal going in and giving you the output is less than a microsecond, which is a very, very short of the time.

Oh yeah, so the late. And yes, be super short, correct? Oh, wow, that's the pain.

Late is a huge, huge thing that you have to deal IT. Right now the biggest source of lencs comes from blue, the way which they are packed SE. And you know .

we been them in fifteen more second on the potential .

innovation. Coal used not uh our final uh wireless communication protocol, everyone to get to highland.

hence the n one in the r one I imagine .

that increases and nx R X .

um yeah that you know the communication protocol blue truth uh allows you to communicate to get father distance .

is more motivation for choosing blue. I mean.

everything has blue th, alright, you can talk to any device.

InterOperability is just absolutely essential, especially in this early face. And in many ways, if you can access the phone or a computer, you can do anything.

We will be interesting to step back and actually look at again, the same pipeline that you mentioned for no one. So what does this whole process look like from finding and selecting a human being to the, to the surgery, to the the first time he's able to use this thing?

So we have was called the patient registry that people can sign up to. Um you hear more about the update and that was about to which newlyn applied. And the process is that once the application comes in, you know IT contains some medical records and we uh you know based on their medical eligibility that there is a lot of different inclusion exclusion criteria for them to me.

And we go through a free screening interview process with someone from the link. And at some point we also go out to their homes to do A P C. I hooted. Um because one of the most kind of revolutionary part about you know having this in one system that is completely wireless is that you can use IT at home.

You don't actually have to go to the lab um and you know go to the clinic to get connector zed to these like specialized equipment that you can take home with you. Um so that's one of the the key elements of you know when we are designing the system that we wanted to keep in mind, like some people hopefully would want to be able to use this every day in the comfort of their home. And um so part of our engagement and and what we're looking for during B, C, A home model is to just can understand their situation, what other technology they .

kind of say that the estimate is uh hundred and eighty thousand people live with quarter piedra in the united states and each year in additional eighteen thousand suffer a paralysing spinal court injury. So laser folks who have a lot of chAllenges living life in terms of accessibility, in terms of doing the things that many of us just take for grant a day today.

And one of the things one of the goals of this initial study is to enable them to have sort of digital autonomy where they buy themselves, can interact with a digital device using just their mind, something that you're calling telaba thy. So digital telepathy, or are a quite gic, can communicate with a digital device in other ways. We've been talking about control the mouse cursor enough to be able to do all kinds of stuff, including play games and tweet, not that kind of stuff. And there's there's a lot of people for whom life, the basics of life are difficult um because of the things that have up to them.

So yeah I mean movement is so so fundamental to our existence, I mean even even speaking involves movement of mouth, live learning and um without that, it's it's it's extremely devolving. Um and there are um there there many, many people that we can help, and I mean especially if you start to kind of a look at other forms of movement disorders um that are not just from special cord injury but from L S A M S or even stroke that that lead you and or just aging right that lead you to lose some of that mobility. That independence is are extremely delicate.

And all of these are opportunities to help people, to help believe IT suffering, to help improve the quality of life. But each other things you mentioned is is only a little puzzle. That needs uh to have increasing levels of capability from a device like in you're link device. And so the first one you you're focusing on is, is just the beautiful word telepathy. So being able to communicate using your mind wirelessly with a digital device, can you just explain this exactly what we're talking about?

Yeah I mean it's exactly that. I mean I I think if you are able to control a uh cursor and able to click um and be able to get access to computer or flown, I mean the the whole world opens up to you and I know I guess the word telephone thy if you kind of think about that as um you know just definitely being able to transfer information from my brain to ea brain um without using some of the the physical faculties that we have you know like voices.

But the interesting thing here is I think I think that's not obviously clear, is how exactly IT works. So in order to move a cursor, there's a at least a couple ways of doing that. So one is you imagine yourself maybe moving a mouse with your hand or you can .

then wish no .

talk about like imagine moving the cursor with your mind like I don't. But it's like there is a cognitive step here that's fascinating because you you have to use the brain and you have to learn how to use the brain and you can't have to figure IT out dynamically like um because you reward yourself if IT works. So you like, I mean, there's a step that this is this just a fascinating step because you have to get brain to start firing in the right way yeah and you do that by imagining like fake, you tell you make IT and always in the right kind of signal that if decoded correctly um can create the kind of effect. And then there's a noise around that if to feature all that up human side, imagine the curse moving is what you have to do.

Yeah, he says using the force of force.

I mean, that isn't that just like, fascinating to you that IT works like to me, like, holy shit, that actually works like you could move a curser with your mind.

you know, as much as your learning to use that thing, that things also learning about you, like our our models constantly updating the weight to say, oh, if if someone is thinking about, you know, this sophistic ted forms of like spiking patterns like that actually means to do this, right?

The machine is learning about the human, and human is learning, so there is a adaptations to the signal process in the decoding step, and then there is the adaptation of known and human being like the same wave. If you give me a new mouse and I move IT, I learned very quickly about its sensitivity. So I learn .

to move IT slower.

And then there's other kind of signal drive that kind of stuff they have to adapt to. Both are adapting to the as a fascinating like soft to a chAllenge on both sides, the software and both on the human software .

and organic orange .

interrogative. Anyway, so so to ruin interrupt. So there's a selection that no one has passed with flying colors. So everything, including that is A, B, C, I friendly home, all of that. So what is the the process of the surgery? Impartation the first moment when he gets to use the system.

the end to end uh you know we say patient to patient out is anywhere between two to four hours uh, in particularly case for no one, IT was about three and a half hours and there is many steps leading to you know the actual robot insertion, right? So there's anesthesia induction and we do intra up CT imaging to make sure that were you know drilling the whole in the right location. And this is also a plan before. And uh someone goes through uh someone like knowledge would go through F M I and then and they can think about weekly their hand and obviously due to their injury, it's not gonna really lead to um any any sort of intended output. But it's the same part of the brain that actually lies up when you're imagining moving your finger to actually moving your finger.

And that's one of the ways in which we can actually know where to place our threads um because we want to go into what's called the hand nap area in the moto cortex and you know as as much as possible densely put our electoral threads um so yeah we do interact up city imaging to make sure and double check the location of the crane I active, my and surgeon comes in, does your thing like skin uh in session crane academic. So drilling of the scholl and then there's many different layers of the brain. There's what's called the dera, which is a very, very thick layer that surrounds the brain, that is actually respective in a process called directory, and that then exposed the P, R.

And the brain that you want to insert. And by the time I spent around anywhere betwen, one, two, one and a half hours robot comes in, doesn't the placement of the target inserting of the thread that takes anywhere between twenty to four minutes, in a particular case, for no one was just under or just over thirty minutes. And then after that the surgeon comes in, there's a couple other steps of like actually inserting the rural substitute player to protect the thread as well as the the brain and then um yeah screw screw in the implant and then skin and then future and .

then you're out.

So when I known walk up was that like was the recovery like and what wants the first time is able .

to use IT? So he was actually immediately after the surgery, like an hour after the church ery as he was waking up. Um we did turn on the device um make sure that we are recording neural signals and we actually did have a couple signals that we um notice that he can actually moderate. And what I mean by module is that he can think about cranching his face and you could see the Spike disappear and appear. That's also and that was immediate, right uh immediate uh after in the recovery room.

how. Yeah, that's a human being. I mean, what would that feel like for you? This device in a human being, a first step of a gigantic journey is a historic moment. Even just to spite, just to be able to module that you know .

obviously there had been other, uh, I mention pioneers that have participated in this ground breaking P C. I. Investigational early feasibility studies. So we're obvious ly standing the shoulders of the giant, not the first ones to actually put electrodes a human human brain.

Um but I mean, just leading up to the surgery there, I definitely could not sleep there, just it's the first time that you are working in a completely new environment. We had a lot of confidence based on our bench top testing, a preclinical R N D studies that the mechanism, the threads, the insertion, all that stuff is is very safe and that is uh no obviously ready for uh, doing this in human. But there's still a lot of unknown unknown about can the needle actually insert.

We brought something like forty needles just in k state break and we ended up using only one. But I mean that that was a level of just complete unknown, right? Because this is a very, very different environment. And that's that's why we do clinical trial on the first place to be able to test these things out. So extreme nervousness and uh just just I many, many sleepless ss night leading up to the surgery and deffand's the day before the surgery and IT wasn't early morning surgery started seven in the morning um and and by the time was around ten thirty IT was IT was IT was everything was done. But I mean first time seeing that.

Well, number one, just just huge relief that this thing is um you know doing what is supposed to do or um and too I just mense among a gratitude for for no man and his family and then many others that have applied and that we've spoken to and will speak to are true pioneers in in every everywhere and you I sort of call them the neural arena or neural net 你 二大爷， you know this is amazing, just in the sixty right is amazing. Just pioneers right um expLoring the unknown outward, in this cases inward um but an incredible amount gratitude for them to uh you know just participate and and play a part um and and it's a it's a journey that were embarking on together um but also like I think IT was just a that was a very, very important muston, but our work was just starting. So a lot of just kind of a anticipation for, okay, s what needs to happen next? Uh, what are set of sequences of events that needs to happen for us to you know make IT worthwhile for um um you know both and .

as just a linger on that, just a huge congratulate to you and the team for that mouse stone. I know there's a lot .

of work well.

but that that that's really exciting to see. There's as a source of hope, this first big step opportunity to help hundreds of thousands of people and there may be a expand the rim of the possible for the human mind for millions of people in the future. So it's really exciting that the the opportunities are all ahead of us. And to do that safely, to do that effectively, was was really fun to see as an engineering watching other engineers come together doing epic thing as awesome.

S, G, S, thank you. Thank you. IT could not have done IT without the team. And yeah, that's the other thing that I I you know told the team as well of just this immense sense of optimism for the future. Um I think I was it's a very important moment for for the company um you know needless to say as well on and hopefully for many others out there that we can all .

the speaking out of chAllenges.

The newer link publish the blog post describing as some of the threats attracted and so the performance as measured by bit per second dropped at first but then eventually was regained and that the whole story of how I was regained this super interesting has definitely something i'll talk to to bliss and to know one about um but in general um can you speak to this whole experience? How is the performance regained and um just the the technical aspects of the threads being attracted, moving. The main .

takeaway is that in the end the performance have come back and the is actually gotten Better than I was before. Um his actually just beat the world record yet again last week um to a point five B P S. So I mean is in a yeah .

the previous world record .

uh in human was four point six yeah so it's a almost and his goals to try to get to ten, which is rough roughly around kind of the media. Neural ink are using a you know mouse with the hand. So it's um is getting there.

So yes, so the performance was regained .

yeah Better than before so that that you know A A A story on its own of what took the B C, I team to recover that performance. IT was was actually mostly on kind of the signal processing. And so now as I mentioned, we were um kind of looking at the Spike outputs from the um our electrode.

And what happened is that kind of a four weeks into the surgery, uh, we noticed that the threads have so they come out of the brain and the way in which we noticed this at first, obviously, is that um I think nolan was the first to notice that his performance was degrading um and I think at the time we were also trying to punch a different experimental um in a different algorithms, different um sort of U I U X. So IT IT was expected that there will be variability in the performance um but we did see kind of a steady decline. And then also the way in which we measure the health of the electrodes or whether they're in the brain or not, is by measuring impedes of the electors.

So we look at kind of the interfacial um of the the handles are good. They say, you know the captain and the and the the resistance between the electro surface and the medium. And if that changes in some dramatic ways, we have some indication or if you're not seeing Spikes on those channels, you have some indications that something happening there. And what we noticed is that looking at those impedance plot and Spike rate plot and also because we have those electrodes recording along the death, you are seeing some sort of movement that indicated that threads were being pulled out.

Um and that obviously, we will have an implication on the model side because if you're the number of inputs that are going to the model is changing because you have less of them on the that model is to get updated, right? And but but the were still signals and as I mentioned, someone to how even when you place the signals on the surface of the brain of the brain or further away outside the school, you see some useful signals. Um what we started looking at is not just the Spike occurrence through this boss algorithm that not mention um but we started looking at just the the the power of the frequency band that is um interesting for no one or no one to be able to module.

So once we kind of change the algorithm, the implant to not just give you the boss output, but also these are Spike band power output. Dad helped us sort we find the model with the new set of inputs and that that was the thing that really ultimately gave us the performance back. Um you know in terms of and obviously like the the thing that we want ultimately and the thing that we are working towards is figuring out ways in which we can keep those threats intact um for as long as possible so that we have many more channels going into the model that's that's by far the number party that the team is current embarking on to understand how to prevent that from happening.

Um the thing that I will say also is that know, as I mentioned, this is the first time ever that were putting these threats in in the human brain and in human brain just for size reference is ten times that out of the monkey brain or the sheep brain and it's um just a very, very different environment. IT moves a lot more, it's like actually moves a lot more than we expected um when we uh did did no surgery and um it's a just a very, very different environment than what were used to. And this is why we do clinical trial, right? We want to uncover some of these uh issues uh and and failure modes earlier than leader.

So in many ways it's provided us with this enormous amount of data and um information to be able to uh solve this. And this is something that newer link is extremely good at once. We have set of clear, objective and enduring problem. We have enormous amount talents across many, many disciplines to be able to come together and fixed the problem very, very quickly.

But that sounds like one of the fascinating Allen es. Here is for the system on the decoding side to be adaptable across different time scale. So whether it's movement of threads or different aspects of signal drift, sort of on the software of the human brain, something changing.

And Donald talks about curr drift, they could be corrected. And there's a whole U. X. ChAllenges how to do that. So sounds like adaptor is like a fundamental property that has to be engineered .

in IT is in and I think I I mean as a company, we're extremely vertically integrated. Um you know we make these inform rays in our own uh, microphone.

Yeah there's like he said, building house. This whole paragraph hear from this blog post is pretty gangster building a technology described above has been no small feet and is a bunch of links here that recommend people click on. We constructed in house micro fabrication capabilities to rapidly produce various situations of thin n film arrays that constitute electoral threads. We created a custom method laser mill manufacture component with microwave l precision. I think there's a tweet associated .

with this whole thing .

that we can get into. Yeah this this okay, what we, what are we looking at here? This thing, this is so in less than one minute, our customer metal second laser meal cuts this geometry in the tips of our needles.

So we're looking at this weirdly shaped needle. The tip is only tend to twelve microns, and with only slightly larger than the the amateurs, a red blood self, the small size allows the us. To be inserted with minimal damage to the cortex. Okay, so what's interesting about this geometry? So we look at this just a different need.

This is the needle that engaging with the loops in the thread. So there are the ones that um you know the loop um and then pa from the silicon backing. And then this is the thing that gets inserted into the tissue and then this pulls out leaving the thread and this kind of a notch or the short tooth that we used to call uh, is the thing that actually is um grasping the loop. And then it's designed in such way, such that you .

pull out robot.

correct? So this is actually house in a canada and basically the robot is has a lot of the optics that look forward. The look is.

There's actually a four or five nanometer light that actually causes the Polly mid to FLorence so that you can locate the the location of the loop. So this light up, yeah, yeah, they do. So micron precision process.

What's interesting about the robot that IT takes to do that? That's that's pretty crazy. That's pretty crazy. That robot is able to get this kind of precision.

Yeah our robot is quite heavy. Um our current version of IT. Um there's I mean it's it's like a giant crsp ve that way is about a ton, because IT needs to be sensitive to a vibration， environmental vibration.

And then as the head is moving at the speed that is moving, you know there's a lot of kind of motion control to make sure that you can achieve that level precision. Um a lot of optics that kind of women on that. Um you know we're working on next generation of the robot that is lighter, easier to transport, I mean is a IT is a feet to move the robot to .

ensure far superior to a human surgeon at this time for this particular task.

absolutely. I mean, let alone you try to actually threw a loop in in a soling kid. I think this is like we're talking like fractions of human hair. These things are not visible. So continue in the paragraph.

We develop novel hardware after testing systems such as are accelerated lifetime testing racks, and simulate the surgery environment, which is pretty al to stress test and validate through a boss of our technologies. We performed many rehearsals of our surgeries to refine our procedures, make them a second nature. This is pretty cool. We practice ges on proxies with all the hardware n instruments needed in our mock or in the engineering space. This helps us the test bastions.

So there's like proxies yeah this proxy super call actually. So there's a three d printed kull from the images that is taken at barrow as well as this uh hydrogel nix sort of synthetic polymer thing that actually mimic the the mechanical properties of the brain. Um IT also has that question of the person. Um so basically what we're talking about here and there's a lot of work that has gone into making this set proxy that um it's about like finding the right concentration of these different synthetic polymers ers to get the right set of consistency for the needle dynamics, you know, as they are being inserted. But we practice this surgery with the person you know's basically physiological and brain are many, many times part to actually doing the surgery to every, every step.

every step.

step. Yeah, like, where does someone stand? Like I like where you're looking at the picture. This is in in our office of this kind of corner of the robot engineering space that we, you know have created this like mock O R space that looks exactly like what they would experience, all the staff word experience during the actual surgery. So it's just kind of like any dense reversal or you know exactly we're onna stand at what point um and you just practice that over over, over, over again with an exact anonyme of someone that you're going to surgery. Ze and and IT got to a point where a lot of our engineers, when we created a crane ec to me that are like that that looks very familiar, which not before .

yeah there is wisdom making game. They're doing the same thing over, over, over. It's like your games, a sushi kind of thing because then it's like olympic athletes visualizes, uh, the olympics. And then once you actually show up, IT feels easy.

IT feels like any other day IT feels almost boring winning the gold metal because you visualized that so many times, you you practice so many times and nothing about us new is boring. You win the gold metal is boring. And the experience they talk about is mostly just relief, probably that they want to visualize IT anymore.

Yeah, the power of the mind to visualize and where I I mean there's a whole field that studies were muscle memory lies in cereBellar yeah it's incredible I think .

is a good place to actually asks of the big question that people might have is how do we know every aspect of this that you describe as safe at end of the day.

the gold standard is look at the tissue um you know what sort of trauma did cause the tissue and does that coral to whatever behavior anomalies that you may have seen um and that's the language to which uh we we can communicate about the safety of you know inserting something into the brain and what type trauma that you can call. So um we actually have an entire department, uh department of pathology that looks at these a teacher slices.

There are many steps that are involved in in doing this. Once you have um you know studies that are launch to with with particular in points in mind, you know at some point you have to use the ize the animal and then you go through necropsy to collect the brain tissue of samples. Um you you fix them in former land and you like gross them, you section them and you look at individual slices just to see what kind of reaction or lack them off excess. So that's the kind of the language to which F D A speaks and you know uh as well for us to evaluate the safety of the insertion mechanism is also the threats um at various different time points, you know both acute um so anywhere between you know a zero to three months to beyond three months.

So those kind of the details of an extremely high standard of safety that has been reached, um F D A supervises this but doesn't general, just a very high standard. And every aspect of this, including the surgery, I think mathematic google is mentioned that like the standard is a let's say, how to put a politely higher than maybe some other Operations that we take for granted. So the the center for all the surgical stuff here is extremely high.

very high. I meat a highly, highly regulated environment with, you know the governing agencies that scrutiny every every medical device that gets marketed. And I think I think it's a good thing um you know it's good to have those high standards and we we try to hold extremely high standards on to kind of understand what sort of damage of any these uh, innovative emerging technologies and new technologies that were building our and you know so far we have been extremely impressed by lack of immune sense from this thread.

Speaking of which you are you talk to me a with the excitement, the histology and some of the images are they are able to share a can you explain to me what .

we're looking at the yeah so what you're looking at is a stained tissue image um so this is a section ed a tissue slice from an animal that was implanted for seven months. So kind of a chronic time point. And you're seeing all these different colors and each color indicates specific type of cell types.

So purple and pink are exercise and micro glia respectably the type of uh clear cells. And ah the other thing that you know people may not be aware of is your brain is not just made up of soup of neurons and axions. There are other uh you know cells like uh geo cells that actually kind of is the glue and also uh react if there any trauma or damage to the tissue.

The Brown are the neurons.

The Brown are the neurons. So so what you're seeing is in in this kind of micro image, you're seeing dislike circle highlighted in White, the insertion site and a when you zoom in to one of those, you see the thread. And then in this particle case, I think we're thing about the sixteen uh, you know wires that are going into the page.

And the incredible thing here is the fact that you have the neurons that are these Brown structures or Brown, circular or electrical thing that are actually touching and a budding the threat. So what is the saying is that there's basically zero trauma that caused during this insertion and with these neural interfaces, these um micro luxuries that you insert, that is one of the most common motor failure. So when you insert these threads like the uti IT causes a Normal death around the site because you're inserting a foreign object, right? And that kind of illicit these like immune response through micro gyan extra sites.

They form this like protective layer around IT h not only are you killing the neuron cells, but you're also creating this protective layer. The then basically prevents you from recording neural signals because you're getting further and further away from the neural unset, your change record and that that is the biggest motor failure. And in this particular example in that inside, it's about fitting micron with the skill bar. The neurons seem to be attracted .

to IT there certainly no beautiful in the way. So the Brown of the neurons for some days I can't look away.

Yeah and the way these things like I mean your tissues generally don't have these beautiful colors on this is a multiplex stain that uses these different uh proteins that are staining. These are different colors. You know that we use very standard set of um staining techniques for h eba one and a new one m and G F.

So if you go to the next image, this is also a kind of illustrates the second point because you can make an argument and initially when we saw the the previous image, we said, oh, like, are the threat just floating? Like what is happening here? Like are you actually looking at the right thing? So what we did is we did another stain, and this is all done in house of the mother's.

Uh, try chrome stain, which is in blue that shows this college layer. So the blue basically like you don't want the blue around the the implant thread because that means that there is some sort of scaring this happen. And what you're seeing, if you look at individual threats that you don't see any other blue, which means that there has been absolutely or very, very minimal to appoint where is not detectable amount of trauma in this insert threats.

So that presume is one of the big benefits of having this kind of flexible thread yeah.

So we think this is primarily to a the size as well as the flexibility of the threads, also the fact that r one is avoiding question. So we're not disrupting or we're not um causing damage to uh the festivals and not breaking any of the the brain berrier a has basically caused the immortals SE to be muted.

But this is also a nice illustration of the size of things. So this is the tip of the thread yeah neons.

There are neurons .

and there this is the thread listening and the electricals a position how yeah so .

this is what you're looking at is not electoral themselves. Those are the conductive wires. so. Each of those should probably be to micron. And with um so what we're looking at we're looking at the coronal slice, so we're looking at uh some slice of the tissue. So as you go deeper, you know you will obviously have less and less of the tapering of the of the thread. Um but yeah the the point basically being that there's just a kind sell around the insert aside, which is I just an incredible thing to see. I've just never seen anything like this.

How easy and safe visit to remove the implant?

Yes, IT depends on when um in the first three months or so after the surgery. Um there there's a lot of kind of tissue modeling that's happening even simple to when you get a cut. Um you know you obviously uh you know start over first couple weeks or depending on the size of ron um scar t 恤 forming right there is like contracted and then in the end they turn IT to scp and you can scare up。 Same thing happens in the brain and it's a very dynamic environment.

And before the scar tissue or the new membrane or you know new members that forms, it's quite easy to just pulled out um and there is minimal trauma that that's a cost during that once the scarred tissue forms. And you know with with knowing as well, we believe that that's the thing that's currently anquan the threads. So we haven't seen any more movements since. So they are they're quite stable um IT IT IT gets harder to actually completely extract the threat. So our current method for a removing the device is cutting the thread, leaving the tissue intact and then unscrewing and taking the in plane and that hole is now going to be plugged with either another neural link or uh just with the know kind of a peak bed you know plastic based cap is okay .

to leave the threads in there forever .

yeah we think so. We've done studies where um you know we left them there. And one of the biggest concerns that we had is like do they migrate and do they get to a point where they should not be? We haven't seen that again once the scar tisa forms 这个 import。 And I should also say that you know, when we say upgrades like it's we're not just talking in theory here like we've actually upgraded many, many times.

A most of our uh monkeys are on human premise. N H, P. Have been upgraded. You know page, who you saw playing mind pong has the latest version of device since two years ago and is seemingly very happy and healthy and fat.

So what's the design for the future, the upgrade procedure. So maybe for and what would the upgrade look like? He was essentially what you're mentioning. Is there way to upgrade sort of the device internally? Will you take IT apart sort of a keep the capsule and upgrade the .

internal couple different things? No one. If we were to upgrade, what we would have to do is um either cut the threads or you know extract the threads depending on kind of you know uh the situation their interns of how the inert garden um if you were to remove them with the rural substitute um you know you have an touch grain so you can we insert different breads on with the updated uh implant package.

Uh there are couple different other are ways that we're thinking about the future of what the upgrade able system looks like. One is you know at the moment we currently remove the dura um this this kind of thick layer that protects the the brain but that actually is the thing that actually prolix ates the scarlet sa formation so typically general good rule of them as you want to leave the the nature as is uh and not disrupt that as much. So we're looking at ways to uh insert the threats through the duo um which comes with different set of chAllenges such as in a pretty thick uh layers.

So how do you actually penetrate that without breaking the needed? So we're looking at different metal design for that as well as the kind of the loop engagement. The other biggest chAllenges are it's quite a pic optically in with White light elimination.

So how do you avoid still the biggest damage that we have of avoiding pascual on how do you image that? How do you actually still mediate other imaging technique we are looking at to enable that? But the goal, the are hypothesis is that, and based on some of the early evidence that we have uh doing through the dr insertion will cause minimal scaring that causes them to be much easier to attract over time.

And the other thing that we're also looking at this is um gonna a fundamental change in the implant architecture is as a at the moment it's a monolithic le implant that comes with the thread that um bond that together so you can actually separate thing out, but you can imagine having two part in plant um you know bottom part that is the thread that are inserted that has the chips um and maybe a video and some power source, and then you have another implant that has more of the computational heavy load and in the bigger battery um and then one can be under the draw. One can be above the draw like you know being the clock for the school, they can talk to each other. But the thing that you want to upgrade the computer and not the thread, if you want to agree that you just go in there, you know, remove the screws and then put in the next version and you know, you know, you know, it's a very, very easy surgery too. Like you do a skin and session slip this in screw, probably be able to do this in ten minutes.

So that would allow you to reuse the threats. Sort of correct? So I mean this least in the natural question of what is the pathworks is scaling the increase in the number of threads. Is that a priority? Is that like what what's the technical chAllenger?

Yeah that that is a party. So for next versions that implant um you know the key metrics that were looking to improve our number of channels just recording from more and more in irons. Um you know we have a pathway to actually go from currently one thousand to you know hopefully three thousand, if not six thousand by end this year and of next year we want to get to um you know even more sixteen thousand there.

A couple limitations to that. One is obviously being able to photos to graphically those wires, as I mentioned, as two micron. And with and in spacing, obviously, there are chiefs that are much more than than those type of resolution.

And we have some other tools that we broaden house to be able to do that. So traces will be narrower, just that you have to have more of the virus coming up into the chips. Chips also .

cannot linearly .

consume more energy as you have more and more channel. So there's a lot of innovations in the circuit. Um you know architecture is as well as a circuit design topology to make them lower power.

Um you need to also think about if you have all of these Spikes, how do you send them off to the end application. So think about then with limitations there and potentially innovations and signal processing. Um physically one of the biggest chAllenges is gonna be a the the the the interface is always the interface that breaks um bonding the stanford array to the um the electronics.

Um IT starts to become very, very highly dense in or connect. So how do you connect tize that there's a lot of innovations um in in of the three the integrations in the recent years that we can take advantage of. Um one of the biggest chAllenges that we do have is forming this hermetic barrier, right? You know this is an extremely harsh environment. They were in the brain, so how do you protected from uh yet like the brain trying out kill your electronics to also your electronics leaking things that you don't want into the brain and that forming that hermetic barrier is gonna a very, very big chAllenge that we and I think are actually ball suit to tackle.

How do you test that? Like what's the development enviro yeah to simulate .

accelerate sentiment is a in of ah IT literally is of that all that is a made up of and again again for all intensive purpose for this particle tech test. Your brain is A A salt and the and you can a also put some other set of chemicals like reactive oxygen es that you know get out of interfaces and cause a reaction to to a pull IT apart. But you could also increase the rate at which these uh interfaces are aging by just increase in temperature.

So every ten degree is that you increase your basically celerity time by two x and there's limit us to how how much temperature you want to increase because at some point there are some other nonlinear dynamics that causes you to have other nasty gases to form that just is not realistic in an environment. So what we do is we increase in our L T, A chAmber by twenty degrees, twenty years. That uh increases aging by four four times. So essentially one day in L T chAmber is four day in calender. And and we look at whether the implants stole are intact, uh, including the threat and an Operation .

or an Operation and all .

of that IT obviously is not an exact same environment as a brain brain has can you know other more biological groups that that attack at IT? Um but IT is a good test environment, testing environment for at least the the the enclosure and the strength of the enclosure. And I mean, we've had implants, the current version of the implant that has been in there for, I mean, close to two and a half years, which is equal on to a decade, and they seem to be fine.

So is interesting that the birth, so basically close approximation is warm salt water. Hot salt water is a good ducting environment. Why do i'm drinking element, which is basically salt water, which is making me kind of IT doesn't have computational power that way the brain does but may be in ms. Of in terms of all the characters, this is quite similar.

Yeah, you've to get them .

in the right P H.

Two, and then consciousness will emerge. Yeah no.

By the the other thing that also is interesting about our enclosure is if if you look at our implant is not your common looking in medical implant, that usually is in the case in the titanium can, that's leasure all. Did we use this polymer copy? F I Polly coral try Floral atlin, which is actually commonly used in blur pack.

So when you have a till and you're try a pop pill, there's a kind of that plastic membrane that's what this is. Um no is actually ever used this uh except us. And the reason we um wanted to do this is because the electromagnet transport. So when we talked about. Electro magnetic inductive charging um with peti um can usually if you want to do something like that, um you know you have have a sfar window and it's a at a very, very tough process to scale.

doing a lot of iteration here. Every aspect of this, the materials, the softer the art.

the whole whole shopping.

So okay, so you mentioned scaling is impossible to have multiple nearing devices as one of the ways of scaling to have multiple nearly .

devices implanted. Um I mean our monkeys have had two newer links, one in each chemistry and that we're also looking at potential of having one in more cortex, one in visual core tex and one in whether cortex. So focusing on .

the particular function, one new link device, I wonder if there is some level of customization that can be done in a compute side. So for the motor core, tex.

absolutely. That's the call you know we talk about at neurally building a generalized neural interface to the brain. And and and that also is strategically how we're approaching this um with with marketing and also you know regulatory, which is hey look um we have the robot and the robot can access any part of the cortex right now we're focused on motor cortex uh with current version of the and one that specialized for motor decoding task but also at the end of the late is kind of a general computer available there um but you know typically if you want to really get down to kind of hyper optimizing for power and efficiency, you do want need to get to have some specialized function right um but what we're saying is, hey, uh you know you you are now used to this robotic insertion techniques which which you took many, many years of showing data um and conversation with fda um and also internally convincing ourselves that is this is safe and um now the difference is if we go to other parts of the brain like visual cortex, which were interested in as our second product um obviously a completely different environment, the court texas laid out very, very differently.

Um you know it's going to be more stimulation focus rather than recording on just kind of creating visual percept. But in the end, we're using the same thin film I technology, we're using the same robot insertion technology. We're using the same, you know, packaging technology. Now it's more of the conversations focus on what are the differences and what are the implication of those differences .

and safety and ethics. Y the way of second product. It's both hilarious and awesome to me, that product being restoring sight for blind people. So can you speak to stimulate the visual cortex? I mean, though the possibilities there are just incredible to be able to give that gift back to people who don't have sight or even in any aspect of that. You can just speak to the chAllenges of the several chAllenges here, many one of which is, like you said, from recording to stimulation, just any aspect of that you've both excited and see the chAllenges of yeah I guess .

i'll start by saying that we actually have been um capable of stimulating through our dental marie as well as as our electronics for years. Um you know we have actually demonstrated some other capabilities for uh reanimating the name in the point of court. You know obviously for for the current E F S study, you know we've hardware disabled that so that that's something that you know you wanted in Spark as a para para tourney um and and you know obviously there many, many different ways to write information into the brain.

The way in which we're doing that is through electrical passing, electrical current and and kind of causing that to really change the local environment so that you can the sort of artificially caused kind of the neurons to depolarizer in in nearby areas for for vision specifically. Um you know the way our visual system works is both well under third, I mean anything what kind of brain the aspects of IT that's well understood but in the end, like we don't really know anything. Um but the way visual system works is that you have full time hitting your eye and in your eyes know there are these specialist cells called photo reduct, their cells that convert the four time energy into electrical signals, and then they get that then gets projected to on your dark of your head, your visual core tex.

Um you know IT goes through actually a clam c system called L G N that then projects IT out and then individual cortex, there's you know visual area one or v one and then there's bunch of other higher level processing layers like v 2v and there there are actually kind of interesting parallels。 And when you study the behavior of convolution neural networks, like what the different layers of the network is detecting, you know first they're detecting like these edges and they are then detecting some more natural occurs and then they started to detect like objects right kind of similar thing happens in the brain um in a lot of that has been inspired and also in been kind of exciting to see some of the correlations there. Um but now things like from there where this cognition arise and where where's color encoded there, there just not a lot of um understanding fundament understanding there.

So in terms of kind of bringing site back to those that are blind, um there are many different forms of blindness. There's actually million people, one million people in the U S. That are legally blind. You know that means like certain uh like score below in kind of the the visual test um I think is something like if you can see something uh at twenty feet distance that Normal people can see at two hundred feet distance like you like if you're all that you legally blind.

that means you can't function effectively using .

site in the world to navigate um and yeah there are the different forms of blinds. There are forms of blindness where uh there is some degeneration of your uh retina, um this photo receptor cells and and rest of your visual uh you know processing that I described tact and for those types of individuals, uh you may not need to maybe stick electrodes into the visual cortex.

You can actually um uh the retinal prosthetic devices that actually just replaces a function of the retina cells that are degenerated. And there are many companies that are working on that, but that that's a very small slice. The all the significance, the smaller slice of focus that are legally blind. Um you if there is any damage along the circuitry, whether it's in the optic nerve or me know H L G N circuit tree or any any break in the circuit that's not going to work for you.

And uh the the source of where you need to actually cause that visual percept to happen because your biological mechanisms are doing that is by placing electronic individual protection the back of your head in the way in which this would work is that you would have an external camera, whether it's a something as unsophisticated as a go pro or you know some sort of wearing able you know riband 他 classes that matters working on that captures the scene right um and that seen a stand converted to set of electrical epulis or stimulation pulses that you would a activate in your visual cortex through um distant from rate。 And by playing some in a concerted kind of orchestra of this stimulation patterns you can create was called fast fans which are these kind of White, yellow wish thought that you can also create by just pressing your eyes. Um you can actually create those percepts by stimulating individual cortex.

And the name of the game is really have many of those, and have those percepts to be the first things be as small as possible so that you can start to tell a part like there are the individual pixel of the the of the screen, right? So if you have many, many of those you know potential, you'll be able to um you know in in the long term be able to actually get naturalistic vision. But in the mid like short term to maybe mid term are being of at least be able to have object detection algorithms run on um on your glasses, uh the prep p processing units and then being at least see the edges of things so you don't bum into stuff.

It's incredible. This is really incredible. So you basically would be adding pixel and your brain would start to figure out what those pixel mean. Yeah and like with the different kinds of the system on the signal crossing on all fronts.

the the thing that actually a couple of things one is um you know obviously you're uh blind from birth um the way brain works, especially in the early age um neural plastic is really nothing other than you know kind of your brain and different parts of your brain fighting for the limited territory yeah yeah I I mean very, very quickly you see you see cases where you know people that are I mean, you also hear about people who are blind that have heightened sense of hearing or some other senses.

And the reason for that is, is that cortex that's not use just gets taken over by these different parts of the cortex. So for those parts of individuals, I mean, I guess they're going to have to now map some other parts of their senses into what they call vision, but it's going to be obviously very, very different conscious experience. Um so I think that's interesting.

carbon. The other thing that also is important to highlights that were currently limited by our biology in terms of the the wavelength that we can see, there is a very, very small wavelength th that is a visible um light wavelengths that we can see with our eyes. But when you have an external camera with this B, C, I system, you're not limited to that.

You can have input, you can have U, V, you can have whatever other spectrum that you want to see. And whether that gets mad to some sort of weird conscious experience, I have no idea. But when often time I talk to people about the goal of nearly being going beyond the limits of our biology and that sort of what I mean.

And if you're able to control the kind of raw signal is that when we use our site with getting the four tons and there's not much processing on IT, if you're be able to control that signal, maybe you can do some kind of processing, maybe you do out to detection ahead of time, yeah, you doing some kind of preprocessing and there's a lot of possibilities, ties to explore that. So it's not just increasing sort of thermal imaging that kind of stuff, but it's also just doing some kind of interesting processing.

I mean, my theory of how like visual system works also is that um I mean, there is so many things happening in the world and there's a lot of full tons that are going into your eye and it's unclear exactly where some of the preprocessing steps are happening. But I I mean I actually think that just just from a fundamental perspective there is so much uh IT. The reality that we're in, if it's a reality um is so there's so much data, and I think humans are just unable to actually like eat enough actually to process all that information. So there are some sort of filtering that does happen, whether that happens in the retina, whether that happens in different layers of the visual cortex, unclear.

But like the analogy that I sometimes think about is, you know if your brain is A C C D camera and the in all of the information in the world there is a sun um and when you try to actually look at the sun with the C C D camera is just gna SATA the sensor because it's enmity amount of energy so you do is you end up adding these filters right to just kind of narrow the information that's coming to you and being captured. And I think you know things like our experiences or are um uh you know like drugs, like profile all that like anesthetics drug or you know psychotic lix. What they're doing is they are kind of swapping out these filters and putting in new ones or removing other ones and kind of controlling our conscious experience.

Yeah, not the distract from the topic, but I just took a very high dose of I walked in the amazon jungle. So yes, it's a nice way to think about IT. You're swapping out different experiences and we're nearly ink being of the control that a primarily at first to improve function, not for entertainment purposes or enjoyment purposes.

but a giving back loss functions.

giving back loss functions and there that that when the functions completely lost, anything is a huge help. Would you imply and your link device in your own brain?

absolutely. I mean, maybe not right now, but absolutely.

What kind of capability once reached, you start getting real? curious. I almost get the anti like like jealous of people that get as you watch them getting planted.

Yeah I mean, I think I mean even even with our early participants, if they start to do things that I I can do um which I think is in the role of possibility for them to be able to get in fifteen twenty if if not like hundred B P S, right um there's nothing that fundamentally stops us from being to achieve that type of performance. Um I mean, I was certainly get jealous that they can do that.

I should say that watching now and I get little jealous because he's saving so much fun and IT seems like such a chill way to play video .

games yeah so I mean to think that also is a hard to appreciate sometimes as that you know he's doing these things while like while talking and means multitasking, right? So it's it's clearly is obviously cognitive intensive, but similar to how know when we talk, we move our hands like these things like you know like our multi tasking, I mean is able to do that. And you know you won't be able to do that with other assistant technology. As far as I am aware know if you're obviously using like an eyes tracking device, you know you're very much fixated on that thing that you're trying to do. And if you're using voice control, I mean, if you say some other stuff.

yeah you don't get to use that. Yeah the the multitask aspect that is really interesting. It's not just the B P S for the primary task is the is the paralyzing of multitask. If you measured the B P S for the entirety of the human organism. So if you're talking and doing a thing with ear, mind and looking around also, but I mean, there's just a lot of paralyzing that they can be having.

I mean, I think at some point for him, but if he wants to really achieve those high level, B, P, S require full attention. And that's a separate section, is a big mystery, like how attention works.

And you know, attention, the cognitive ad of literature, people doing two tasks like, uh, you have your primary task in the secondary task. And the secondary task is is a source of distraction.

And how does that affect the performance of the primary task? And there is depending on the tasks, there is a lot of interesting, I mean, this is an interesting computational device, right in, I think there, to say the least, a lot of novel insights that can be gained for every any I person surprised, I know, is able to do such incredible control, the curse or wild king, and also being nervous at the same time, because he's talking like all of us are. If you're talking in in front of the camera, you get nerves.

So all of those are coming into playing, able to still eve, high performance, surprising. I mean, all this is really amazing. And I think just after researching this really in depth, I can't want in your ink .

IT get in line. And also the safety, get mine.

Well, we should say the registry is for people who have quite political. And ca, so theyll be a separate line for people. They're just curious, like myself, so now that no one patient p one is part of the ongoing to study, um what's the high level vision for p two, p three, P P five and just a the expansion into other human being there getting to experience this .

implant yeah I mean, the primary goal is you for for our study in the first place is to achieve safety and points to understand safety of this device as well the implantation process um and also at the same time understand the efficient y and the impact that he could have on the potential users lives on and.

Just because you have and you know you're living with tetraplegic IT doesn't mean your situation as same as another person living with tech pledge is widely, widely vine and and you know you something that you know we're hoping to also understand how our technology can serve not just a very small slice of those individuals, but you know broader group of individuals. And being able to get the feedback to you know just really build the just the best product for them. Um so are are you know there there's obviously also uh you know goals that we have.

And in the primary purpose of the early feasibility study is to learn from each and every participants to improve the device, improve the surgery before you know we embark on what's called the pivotal study that is a much larger um trial that starts to against statistic significance of your n points um and that's require before you can then market the device. Uh and you know that's how IT works in the U S. And just generally around the world the process you follow. So in our our goal is to really just understand from people like knowin, p two, p three future participants, what aspects of our device needs to improve. You know, if if IT turns out that people like I really don't like the fact that at last only six hours, I want to I be able to use this computer for, you know, like twenty four hours, I mean that that is up, you know, user needs and user requirements on which we can only find out from just just being able to engage with them.

So before the pivotal study, there's kind of like a rapid innovation based on individual experiences. You are learning from individual people how they use IT like the like highly olustee ils in terms of like Chris control and signal like stuff to life experience.

So there is hardware changes, but also just just former updates. So even even when we you know had had that sort of a recovery event for no one, you know he now has a new format that that he um has been uh updated with and know someone to how like your phone get updated all the time with new farmers for security patches, whatever new functionality.

U I right um and that's something that is possible with our implant, is not a static one time device that that can only do the thing that IT said I can do. I need somewhere to tesla la, you can do over the air former update. And now you have completely new ur user interface um all this bells and whistles and improvements on you know everything like the latest right H A latter were talking .

about the up. No one is using there's like cellular tion, all that kind of stuff. And then this update just you just clicking get update.

Uh, what other future capabilities are? Are you kind of looking to, he said, vision? That's a fast.

Anyone 啊 what about sort of accelerated typing or speech， this kind of stuff? yeah. And what else is there .

are those are still in the room volume of movement program. So it's a largely speaking, we have two programs. We have the movement programme and we have the the vision program.

The movement program currently focus around you know， the digital freedom. As you can easily guess, if you can control, you know two cursor in the digital space, you could move anything in the physical space. Um so robotic arms, we'll chair your environment uh or even really like whether is through the phone or just click directly to those interfaces. So like to those machines.

Um so we're looking at ways to to expand those type of capability even for knowledge um that requires in a conversation with the F D A and of showing safety data, there's a robotic guard, more real chair that you know we can guarantee that they're not gonna t themselves and is very different if you're moving stuff in in the digital sal space. Um potentially caused harm to the participants. Um so were working through that right now.

Um speech does involve different areas of the brain. Speech pathetic is very, very fascinating. And there's actually been a lot of really an amazing work that's been happening in academia, in a study at U C. Davis, Jamie henderson son and in a late Christian oy um as stanford are doing to some incredible amount work and improving speech, uh neo protheus s and those are actually looking more at parts of more or cortex that are controlling in these vocal and you know being able to like even by mouthing the world or imagine speech, you can pick up those signals. Um the more sophisticated higher level processing areas like you know the progress area or you know a warning is those are still very, very big mystery inference of the the underlying mechanism of how all that stuff works. But um yeah I think I think nearly eventual goal is to kind of understand those those things um and and be able to provide a platform and tools to be able to understand that and study that.

This is where I get to the pot head questions um do you think we can start getting insight into things like thought? So speech is a there's a moscow components like he said, there's like the act of producing sounds. But then what about the internal things like cognition, like low level thoughts and high level thoughts? Do you think we'll start noticing kind of signals that could be picked up, that they could be understood, that could may be used in order to interact with the outside world .

in some ways? Like, I guess, this starts to come again until the hard problem of consciousness um and um. I mean on on one hand all of these are at some point set of electrical signals that um from there maybe IT IT in itself is giving you the cognition or the meaning or somehow human mind is incredibly amazing story telling machine so we're telling ourselves and falling ourselves that there are some interesting meaning here um but I I certainly think that P C I you know really P C ended.

There is set of tools that help you kind of study the underlying mechanisms and in both like local but also broader sense um and whether you know there are some interesting patterns of like electrical signal, that means like you're thinking this verses and you can either like learn from like many, many sets of data to correlate some of that and be able to do mine reading or not. I'm not sure um I certainly would not kind of bull that out as a possibility, but um I think B C, I alone probably can do that. There's probably additional set of tools and framework again. And also like this hard problem of consciousness at the end of the day, is rooted in this for sophy question of what is was the meaning of IT all the nature of our existence? Like, where's the mind emerge from this complex network?

Like, yeah, how does the how does the subjective experience emerge from just a bunch of Spikes?

Electrical Spikes? Yeah yeah mean, we do really about B, C, I. And we're building as a tool for understanding the mind, the brain. The only question .

that matters .

there's actually um there actually is um some biological the existence proof of like what you would take to kind of start to form some of these experiences that may be unique. Um if you actually look at every one of our brains there, there two hemispheres, there's a left side of brain, there's a right side of brain.

And I mean I unless you have some other conditions, you Normally don't feel like left left or right legs like you just feel like one legs, right. So what is happening there, right? Um if you actually look at the .

two .

hemispheres, there's a structure that can connect ze, the two called the corporate colossus that is supposed to have around two hundred, three hundred million connections or axions um so what that means that the the number of interface and electrons that we need to create some sort of mind mild or from that like whatever new conscious experience that you you can experience. Um but I do think that there is like kind of an interesting existence proof that we all have and .

that threshold is unknown at this time. Oh.

these things everything in this domain is in a speculation, right?

And there will be a you be continuously pleasant surprise. Do you see a world where there's millions of people like tens of millions, hundreds of million that people walking around with the neural link device and there are multiple neural devices in .

their in I do first all there there are like if you look at worldwide um people suffering from movement disorders, visual deficits 上面 uh in the tens of not hundreds of millions of people um so that alone， I think there's a lot of uh benefit and and potential good that we can do with this type of technology. And when you start to get into connect neural like psychiatric application, no depression, um anxiety, hunger or in obesity, right like mood control of appetite, I mean that starts to become you know very real to everyone.

Not to mention that every uh most people, others have a smart phone. And once B, C, I starts competing with a smart phone as a preferred methodology of interacting with the digital world.

that also becomes an interesting thing. Oh yeah, I mean that yeah, this is even before going to that, right? I mean, there is like almost, I mean, the entire world that could benefit from these type of thing.

And then yeah like if you're talking about kind of next generation of how we interface with no machines or even ourselves. A in many ways I think um I can play a role in that. Um and you know some of the things that I also talk about this, I do think that there is a real possibility that you see eight billion people walking around with your link. Well.

thank you so much for pushing ahead, and I look forward to that exciting future. Thanks for me. Thanks for listening to this conversation with DJ. So and now, dear friends, here's Matthew mcdougle, the head near a surgeon at near link. When did you first become fascinated with the human brain .

since forever? As far back as I can remember, i've been interested in the human brain. I mean, I was, you know a thoughtful kid and a bit of an outsider and you you know sit they're thinking about what the most important things in the world are in your in your little tiny adolescent brain and the answer that I came to, that I converge, don, was ah that all of the things you can possibly conceive of as things that are important for human beings to care about are literally contained. You in the kull, both the perception of them and their relative values. And you know, the solutions, all our problems and all of our problems are all contained in the skull. And if we knew more about how that worked .

and how the .

brain encodes information and generates desires and generates agony and suffering, we we could do more about IT. You know, you think about all the all the really great triumph in human history, you think about all the really horrific tragedies, um you know, you think about the holidays, do you think about um any prison full of human stories? And all of those problems boiled down to neurochemistry.

So if you get a little bit of control over that, you provide people the option to do Better. In the way I read history, the way people have dealt with having Better tools is that they most often, in the end, do Better with huge s risks. But I think it's a and interesting, a worthy and noble pursuit to give people more options, more tools.

Yeah, that's a fascinating way to look at human history. You just imagine all these neurobiological mechanism stalled and hit. Learn all of these jank s kon. All of them just had like A A brain. He just a bunching in neurons, you know, like a few times of billions of neurons are gaining a bunch information over a period of time.

They have set a module that this language of memory in all that and from there, in in the, in the case of those people, they are able to murder millions of people here and and all that coming from. There's not some glorified notion of A A dictator of this enormous mine or something. This is just, is just the brain.

Yeah, yeah. I mean, a lot of that has to do with how well people like that can organize those around them. Other brains, yeah. And so I always find IT interesting to look to primatology.

And I looked to our closest nonhuman relatives for clues as to how humans are going to behave and and what particular humans are able to achieve. And so you look at chimpanzees and monopoles, and similar, but different in their social structures particularly. And I went to emi in ata, studied under friends to all the great friends to all who was kind of the leading primatologist. I who recently died, and is working at looking at champs through the lens of, you know, how you would watch an episode of friends and understand the motivations of the characters are acting with each other. He would look at a chip colony and basically apply that lends a massively over simplifying IT. If you do that, instead of just saying you subject for seven, three, you know through his faces at subject for seven one, you talk about them in terms of their human struggles, accord them the dignity of themselves as actors with understandable goals and drives what they want out of life and primarily to know the things we want out of life food, sex, companionship, um power. You can understand chip and panopea behavior in the same lights much more easily and I think doing so gives you the tools you need to reduce human behavior from that kind of false complexity that we layer onto IT with language and look at IT in terms of all these humans are looking for companionship, sex, food, power um and I think that that's a pretty powerful tool to have an understanding human behavior.

And I just went there on john goal for a few weeks and it's a very visceral reminder that a lot of life on earth is just trying to get laid yeah, they're all screaming at each other. I I saw a lot of monkeys and they're just trying to impress each other. Or maybe there's a battle for power, but a lot of the battle of the power has do than getting laid.

Reading rights often go with all the status. And so if you can get a piece of that, then you can na do O, K. And I would like to think .

they were somehow fundamental different. But especially when you talk comes to prime, is what really aren't you know, we can use fancy poetic language, but maybe some of the underlying drives and motivators are similar.

Yeah.

that's true. And all that is coming from this, the brain. Ah, so when did you first start starting the brain? Is because the biological mechanism.

basically the moment I got to college started looking around for labs that I could uh do neuroscience work in. Uh I originally approach that from the angle of uh, looking at interactions between the brain and the immune system, which is in the most obvious place to start.

But um I had this idea at the time that the contents of your thoughts would have an impact, a direct impact, maybe a powerful one on non conscious systems in your body, the systems we think of, as you know, homeostatic automatic mechanisms like fighting off a virus, like repair, ring a wound um and turn off. There are big crossovers between the two. I mean, IT gets to a kind of a key point that I think was under recognize one of the things people don't recognize or or appreciate about the human brain enough. And that is that IT basically controls or has a huge role in almost everything that your body does um like you tried tried a name, an example of something in your body that isn't directly controlled or massively influenced by the brain. And um it's pretty hard, I mean might say like bone healing or something but even those systems the hypotheses sympathy ati end up playing a role in coordinating the indecent system that does have a direct influence on, say, the calcium level in your blood that goes to bone healing so non obvious connections between those things implicate the brain is really a potent prime mover in all of health.

One of things I realized in the other direction to how most of the systems of the body integrated with the human brain, like they affect the brain also like the immo system. I think there's just you know people who study all zimba and those kinds of things is is surprising. How much you can understand of that from the immune system, from the other systems that don't obviously, they seem did any do with of the nervous system, they all play together. Yeah.

you could understand how that would be driven by evolution to just in some simple examples, if you get sick, if you get a community disease, you get the flu. Uh, it's pretty advantages for your immune system to tell your brain aid. Now, be anti social for, you know, a few days.

Don't go be the life of the party tonight. In fact, maybe just cut up somewhere warm under a blanket and just stay there for a day or two. And sure enough, that tends to be the behavior that you see both in animals and in humans. If you get sick, elevated levels of interactions in your blood and tnf, alf and your blood, ask the brain to cut back on social activity and uh, even moving around the lower look, motor activity in animals, their infected viruses.

So from there, the early days in your science to surgery, when did that stuff happen? Yes, leap.

I was sort of an evolution of thought. I wanted to study the brain. I started studying the brain, uh, in underground, in this neo imminent logy lab.

Uh, I from there realized at some point that I didn't want to just generate knowledge. I wanted to affect real changes in the actual world, in actual people's lives. And so after having not really thought about going in the medical school, I was on a track to go to a PHD program.

I said, well, i'd like, i'd like that option. I'd like to actually potentially help tangible people in front of me. And doing a little digging, found that there exists is M D P H D programs, or you can choose not to choose between them and do both. And so, uh, I went to usc for medical school and had a joint P H D program with caltech, where I met a try, chose that program, particularly because of a researcher at caltech named Richard Anderson, who is one of the godfathers of primate neuroscience, has A A macc lab where uta ys and other electrodes were being inserted into the brains of monkeys to try to understand how intentions were being encoded in the brain.

So I know I ended up there with the idea that maybe I would be a neurologist and study the brain on the side and then discovered that neurology again, i'm going to make enemies by setting neurology predominated and and pressingly to me is is the practice of diagnosing a thing and then saying good luck with that when there is not much we can do um and a surgery very differently uh is a is a powerful lever on taking people that are headed in a bad direction and changing their course uh in the sense of brain tumors that are potentially treatable or curable with surgery. Um you know even animals in the brain, blood vessels that are going to capture you save lives, really is at the end of the day, what what mattered to me. And so um I was at U.

S. C, as I mention that happens, be one of the great neurosurgery programs. And so I met these truly epic, uh, neurosurgeons, uh, alex classy and and Michael zz le and Steve gian out and martial SE sort of epic people that, or is human beings in front of me. And so IT kind of change my thinking from neurosurgeons are distant gods that live on another planet and occasionally come and visit us to these are humans that have problems in our people and uh there's nothing fundamentally preventing me from being one of them uh and so um at the last minute in medical school, I changed gears from going to do different specialty and switched in the neurosurgery uh, which cost me a year, had to do another year of research ah because I was so far along the process that to switch in the neurosurgery deadlines that already passed a decision that cost time but absolutely worth IT.

What was the hardest part of the training on the newer surgeon track? Yeah two things .

I think that in a residency in your surgery sort of a competition of pain of like how much pain can .

you eat and smile .

yeah and so there's work out restrictions that are not really their views that I think internally among the residents as weakness. And so most neurons gey residents try to work as hard as they can and that I think necessarily means working long hours and sometimes over the work. Our limits and you know we care about being compliant with whatever regulations are uh in front of us. But I think more important than that, people want give all give their all in becoming a Better neurosurgeon because the the stakes are so high and so it's a real fight to get residence uh to say go home at the end of their shift and stay and do more surge. Are you .

seriously saying like one of the hardest things, literally like getting forcing them to get sleep and rest in all this castle?

Historically, that was the case. I think I think next generation, I think the next generation is more a compliant .

and more secret. What mean? I didn't say that.

Now i'm making enemies. No, okay, I get IT. Wow, that's fascinating. So what was the second thing?

The personalities and maybe the two are connected.

but so is that was a pretty competitive .

competitive and it's also, um you know as we touched on earlier, primates like power. And I think neurosurgery has long had this ora of a mistake and excEllence and whatever about IT.

And so it's it's an invitation, I think for people that are locked in that authority, you know board certified our surgeons basically a walking falaise appeal to authority, right? You you have license to walk into any room and act like your, you know, an expert on whatever. And fighting that tendency is not something that most neurosurgeons do. Well, humility isn't the .

fourth yeah one of the so I friends uh who know you and whenever they speak about you that you you have the surprising quality for a neurosurgeon in of humility. I think in the case it's not it's not as communist, perhaps other professions because there is a kind of gigantic sort of heroic aspect of neurosurgery. And I think IT gets to people's had a little .

bit yeah well that I think that ah you know that allows me to play well at an elan because ilan one of his strength I think is just instantly see through fala acy from authority so nobody walks into a room that his finances will god damage. You have to trust me. I'm the guy that built the last, you know, ten rockets or something.

He is what you did IT wrong, and we can do a Better. Or I am the guy that you kept forward alive for the last fifty years. You listen to me on how to build cars and he says, no and so you don't walk into a room, the east end, and say, well, i'm a neuro surge.

Let me tell you how to do IT. I is gonna. Well, i'm a human being that has a brain I can think from first principles myself thank you very much ah and here's how I think I had to be done. Let's go try and see who's right uh and not to you know proven, I think over and over in his case.

to be a very powerful approach. We just take that engine. There's a fascinating team at your a link that you get interact with, including you on what do you think is the secret to a successful team? What have you learned from just going to observe these folks? Yeah world experts in different disciplines work together.

Yeah, there there's a sweet spot where people disagree .

and .

forcefully speak their mind and passionate, defend their position, and yet are still able to accept information from others and change their ideas when they're wrong. And so I like the analogy of sort of .

how you polish rocks.

You put hard things in a in a hard container and spin IT people dash against each other and outcomes, uh, you know, a more refined product. And so to make a good team at our link, we try to find you know, people that are not afraid to defend their ideas passionately and you know occasionally strongly disagree with people that that they're working with and have the best idea come out on top.

It's not an easy baLance again to refer back to the primate brain. It's not something that is inherently built into the the primate brain to say I passionately put all my chips on this position and now i'm just going to walk away from and admit you are right you part of our brains tell us that that is a power loss. That is a loss of face, loss of standing in the community and uh and and now you're A A data jump because your idea I got tranced um and you just have to you know recognize that that little voice in the back, your head is more adaptive and not helping the team win yeah I have .

the confidence you to walk away from an idea that you hold on to ah yeah and if you do that often enough, you're actually going to become the best food in the world you're thing I mean that kind of that rapid .

oration yeah you at least be a member .

of a winning team right the way ah what what did you learn? You mention there's a lot of amazing a new assurances that U S C, what what listens about surgery in life have you learn from those books?

Yeah I think working your ask of working hard well um you know functioning as a member of a team, getting a job done that is incredibly difficult um you at working incredibly long hours, being up all night, taking care of someone that ah you know you think probably we won't survive and no matter what you do, working hard to make people that you passionate dislike look good .

the next .

morning these folks were relentless in their pursuit of um excEllent neurosurgical technique decade over decade and and I think were well recognize for the that excEllent so special martial SE Steve gian, not a Michael zo. They made huge contributions not only to surgical technique but they built training programs that trained dozens or hundreds of amazing their a surgeons I was just lucky up got to be in their week.

What's that like imagined doing a surgery where the person is likely not to survive? Is that where I knew?

yeah.

You know, IT is especially .

chAllenging .

when you with all respect to our elders, IT doesn't hit so much when you're taking care of an eighty year old and something was going to get them pretty soon anyway. And so you lose a patient like that. And IT IT was part of the natural course of what is expected of them in, in the coming years regardless. A taking care of, you know, a father of two or three four Young kids, someone in their thirties that didn't have IT coming and they show up in your E R, having their first seizure of their life. And long old, you've got A A huge moignon inOperable or incurable brain tumor.

You you can only do that I think a handful of times um before really starts eating away your at your armor sorry, you know a Young mother that shows up that has a giant hammer gender brain that she's not onna survive from you know they bring her four year old daughter and to to say goodby e one last time before they turned event later off that. You know the great Henry marsh is an english neurosurgeon who said at best I think he says every neurosurgeon Carries with them a private graveyard and I definitely feel that um especially with the Young parents that that kills me. They they had a lot more to give the the loss of those people specifically has a you know knock on effect that's going to make the world worse for people up for a long time.

And it's just hard to feel power less in the face of that. You know, that's where I think you have to be a borderline evil to fight against a company like neurology. Tip consistently be taking potshots at us because what we're doing is to try to fix that stuff.

We're trying to give people options, uh, to reduce suffering. We're trying to we're trying to take the the pain out of life that broken brains brings in. And yeah, this is just our our little way that we're fighting back against entropy. I guess yeah.

that's the the amount of suffering is endured when some of the things that we take for granted that our brains able to do taken away is immense and to be able to restore some of the functionality is a real gift.

Yeah, we're just starting where we're we're going to do so much more.

Um well, can you take me through the full procedure implanting, say the one sure chip and link .

yeah it's a really simple, really simple straight ford procedure. The human part of the surgery that I do is dead simples. One of the most basic neurosurgical procedures imaginable. And I think there is evidence that some version of IT has been done for thousands of years of the examples I think for me egypt of healed or partially healed ah refinance and from you know ancient times in south amErica where ah these proto surgeons would drill holes in people's skills, you know presumably to let out the evil spirits but maybe to drain blood clots and there's evidence of bone healing around the edge, meaning the people at least survive some months uh after a procedure.

And so what we're doing is that we are making a cut in the skin on the top of the head over the area of the brain that is the most potent, uh, representation of hand intentions. And so if you if you are an expert counter pianist, you know, this party of brain is lighting up the entire time you're playing. We call the hand nob, the hand nob.

Yes, all the like, the finger movement, all this, all, all of that is just fire and away. Yes, there's a little .

squirrel in the cortex right there. One of the folds in the brain is kind of double folded right on that spot. So you can look at IT on A M R.

I say, that's the hand up. And then you you do a functional test and a special kind of M R I and a functional M R I, F M R I. And this part of the brain lights up when people are even quite a pleasant people whose brains aren't connected to their finger movements anymore. They imagine finger movements in this part of the brain still lights up.

So we can I D that part of the brain in anyone who's preparing to enter our trial and say, okay, that part of the brain we confirm is your hand intention area um and so i'll make a little cut in the skin, will flap the skin just like kind of opening hood of a car, only a lot smaller. Make a perfectly round one in to diameter hole in the skull, remove that bit skull, a open the lining of the brain, the covering of the brain. It's like a, like a little bag of water that the brain float in, and then show that part of the brain to a robot.

Then this were the robot shines that can come in and take these tiny, you know, much smaller than human hair electrodes, and precisely insert them into the cortex, into the surface of the brain, to a very precise depth, in a very precise spot that avoids all the blood vessels that are coating the surface of the brain. And after the robots done with its part, and you know, the human comes back in and puts the implant into that hole in the skull and covers IT up, screwing IT down to the skull and showing the skin back together, that the whole thing is, you know, a few hours long, it's extremely low risk compared to the average near surgery involving the brain that that might say, open up a deep part of the brain or manipulate blood vessels in the brain. This this opening on the surface of the brain with with only critical micon sessions Carries significantly less risk than a lot of the, you know, tumor or analisa surgeries that are routinely done.

The critical microinsurance that are V A robot and computer vision are designed to avoid the blood vessels. exactly. So I know your bit bye is here, but let's compare human machine. So what are human surgeons able to do well? And what are robot certains to do well at the stage of our human solution development?

Yeah, yeah. It's good question. Humans, uh, are general purpose machines, were able to adapt to unusual situations, were able to change the plan on the fly. I remember well a surgery that I was doing many years ago, down in Sandy ago, where the plan was to open a small hole behind the year and go reposition of blood vessel that had come to lay on the facial nerve, trigonal nerve, the nerve that goes to the face when that blood vessel lays on the nerve, IT can cause just intolerable, horrific shooting pain. The people described like things up to the cattle prod.

And so the beautiful, elegant surgeries to go move this blood vessel off the of the nerve, the surgery team, we we went in there and started moving this blood vesle and then found that there was a giant anisa on that blood vessel that was not easily visible on the prefs scans. And so the plan had a dynamically change, and that the human surgeons had no problem with that were trained for all those things. Robots wouldn't do so well in that situation, at least in their current in oration fully robotic surgery like you know, the the electorate and service portion of of the neural ink surgery, IT goes according to a set plan.

And so the humans can interrupt the flow and change the plan, but the robot can really change the plan midway through. IT Operates according to how I was programmed and how I was asked to run. IT does its job very precisely, but not with a wide degree of attitude and how to react to changing conditions.

So there could be just a very large number of ways. You could be surprises, a surgeon when you enter a situation that could be yet subtle things that after dynamic, just after correct. And robots are not good at that.

Currently renting. I think we are the dawn of a new era with A I of the parameters for robot responsibility to be dramatically broadening, right? I mean, you can't look at a self driving car and say that it's Operating under very narrow parameters.

You know, if a chicken runs across the road, IT wasn't necessarily programmed to deal with that specifically, but IT a way more a self driving tesla. We have no problem reacting to that appropriately. Uh, and so surgical robots aren't there yet, but give IT time.

And then there could be a lot of into IT, like simulating omas possibilities of maybe robotic surging. Could say this situation is perfectly familiar, or the situation is not familiar. And in the not familiar case, human could take over.

I basically like, be very conservative and saying, okay, this for sure has no issues, no surprises, and let the humans deal with the surprise, with the education, all that, that's one possibility. So you think eventually, uh, you be out of the job. What you being your surgeon, your job being your surge. Humans, there will not be many new surgeons .

left on the earth. I'm not worried about my job in my in the course of my professional life. I think I would tell my my kids not necessarily to go in this line of work depending on depending on how things look in twenty years.

It's so fast. And because I I mean, I if I have a line to work out, says programing and if you ask me, like you ask I don't know twenty years what I would recommend for people, I would I would tell me, yeah, go you will always have a job with your programmer. There's more, more computers and all this kind of IT pays well.

But then you realize these largest language models come along, and they really did have good at generating code. So overnight you could be surprised, what is the contribution of the human, really? But then you start to think, okay, IT does seem that humans have ability, like you said, to deal with novel situations.

In the case of programming, it's the ability to come come up with novel ideas to solve problems. So IT IT seems like machines aren't quite yet able to do that. And when mistakes are very high, when its life critical as IT is in surgery, especially newer surgery, then he starts the the stakes are very high for robot tax are replacing human, but is fascinating. This case of neural link, there's a, uh, human robot collaboration.

Yeah, yeah, it's I do the parts I can do and what does the parts I can't do. And we we are friends.

That I I saw that there's a lot of practice going on. So I mean, everything that you like is is tested extremely rigorously. But one of things that I saw that there is a proxy on which surgeries are performed yeah. So this is both for the robot and for the human, for everybody involved in the entire pipeline. What's that like practicing .

the surgery? It's pretty intense. Uh, so there is no analog to this in human surgery. A human surgery are of this artisan craft that handed down directly from master to pupil over the generations, is, I mean, literally, the way you learn to be a surgeon on humans is by doing surgery on humans. Mean, first you watch a.

your professors .

do a bunch of surgery, and then finally they put, know, the trivial parts of the surgery and your hands and then more the more complex parts. And as you're understanding of the in the point in the purposes of the surgery increases, you get more responsibility in a perfect condition has always go well. In new length case, the approaches is a bit different.

We, of course, practice as far as we could on animals. We did hundreds of animal surgeries um and when I came time to do the first human um we had a amazing team of engineers build incredibly life like models. One of the engineers friend remond m in particular, built built a pulsating brain in a custom three printed skull that matches exactly the the patients anatomy, including their face and scale characteristics.

And so when I was able to practice that, I mean, that is as close as I really reasonable should get to to being the real thing in all the details, and including, you know, the having a american body attached to this custom head. And so when we were doing the practice surgeries with wheel that body into the city scanner and take a mock C T. Scans and wheel back in and conduct all the Normal safety checks verbally, you stop.

This patient were confirming his identification is manica number, blow, blow, blah. And then opening the brain and exactly the right by using standard Operative neun navigation equipment, standard surgical drills, and in the same awar that we do all of our practice surgeries, and at neural ink, and having the skull open and have the brain pulls, which adds a gree of difficulty for the robot, you know, perfectly precisely, plan and insert those electrodes to the right depth in location. And so yeah, we we kind of broke new ground on how extensively we practiced for the .

surgery so there was a historic moment, a big milestone for your link in part for humanity with the the first human getting a neural and plant in january of this year. Take me through the surgery unknowns what do you feel like to be part of this?

Yeah well we um we're lucky to have just incredible partners at the very neurologic institute. Yer I think the premier neurosurgical hospital in the world, they they made everything as easy as possible for the trial to get going and and help the immensely with their expertise on how to how to arrange the details. IT was a much more high pressure surgery in some ways.

I mean, even even though the you know, the outcome wasn't particularly in question in terms of our participant's safety, the number of observers, you know the number of people, there is comfort terms full of people watching live streams in the hospital um rooting for this to go perfectly. And that just adds pressure that is not typical for even the most intense production neurosurgery, say removing a tumor or you know placing deef brain stimulation electrodes and IT had never been done on a human before. There were unknown, unknown.

And so a definitely a moderate pocky factor there for the whole team, not knowing if we are going to encounter, say, uh, a degree of brain movement that was unanticipated, or a degree of brain sag that took the brain far away from the skull, made IT difficult to insert, or some other unknown, unknown problem. Fortunate everything. I went well, and that that surgery is one of the smoothest outcomes that we could imagined.

We knows, I mean, extremely back. And like in the superboss of situation.

tremeau extremely. I was very pleased when I went well and when I was over. Looking forward to number two.

Yeah even with all that practice, all of that just we've never been in a situation that's so high stakes in terms of people watching yeah and we should also probably mention, given how the media works, a lot of people, you know maybe in a dark kind away hoping he doesn't go well well that .

I think wealth is easy to hate um or envy or whatever. And I think this whole industry around driving clicks and bad news is great for clicks. And so any way to take an event and turn IT into bad news ah is going to be really good for for clicks.

IT just sucks because they think in IT puts pressure on people IT discouragedly people from from trying to solve really hard problems. Because to solve hard problems have to go to the unknown. They have to do things we've been done before, and they have to take risks, calculated risks. You have to do all kinds city precautions, but risks neural lesson ah, I just wish you would be more celebration of that, of the rest taking verses like people just waiting on the on sidelines, like waiting for failure yeah and then pointing out the failure yeah sucks. But no, in this case it's it's really great that everything was just floor ssl, but is unnecessary pressure.

I was a now that there is a human with literal skin in the game, there is a participant who whose well being rides on this doing well. You have to be a pretty bad person to be rooting for that to go wrong um and so you know hopefully people look in the mirror and realized that at some point.

So did you get to actually front row seat like wash the robot work like what uh get to see the .

whole thing yeah I mean you know because an M D needs to be in charge of all of the medical decision making throughout the process. Um he unscrupulous the surgery after exposing the brain, presenting IT to the robot and um place the targets on the robot um into software interface that tells the robot where it's going to insert each thread that was done um with you know my hand on the mouse for whatever that's .

worth so you were on place in the targets oh cool so I get you know the robot uh what the computer vision provides a much candidates and you to finalize the decision .

you know the the the softer engineers are amazing on this team and so they actually provided an interface where you can essentially use a last so tool and select A A prime area of brain real estate. And IT will automatically avoid the blood vessels in that region and automatically place unit targets. See, know, that allows, you know, the human robot Operator to select a really good areas of brain and make dense applications of targets in that in those regions, the regions we think going to have the most um high fidelity representations of finger movements and ARM movement intentions.

I've seen the images of this in for me with O C D for some reason are really pleasant. I think there's a sabbatic called oddly .

satisfying yeah love that it's .

oddly satisfying to see the different target size avoiding the blood vessels and are also maxims ing like the usefulness of those locations for the signal. He just feels good as as a person .

who has a visual reaction to the brain bleeding.

I can tell you, yes.

it's extremely satisfying watching the electrodes themselves go into the .

brain and not cause bleeding yeah ah so you said the feeling was a relief when everything went perfectly. Now how deep in the brain can you currently go and eventually go to the new links side? IT seems the deeper is on the brain, the more chAllenging IT becomes.

Yeah so talking broadly about neurosurgery, we can get anywhere. It's routine for me to put deep brain stimulating electrodes, uh, near the very bottom of the brain, a entering from the top and passing, and about a two millimeter wire all the way end to the bottom of the brain. And that's not revolutionary.

A lot of people do that, and we can do that with very high precision. I I use a robot from global do that surgery um you know several times a month. It's it's pretty routine.

What are your eyes in that situation? What what are you seeing? What's what kind of technology can use to visualize where you are to delight you way? yes.

So it's cool process on the software side. You take a preOperative M R I that's extremely high resolution date of the entire brain. You put the patient to sleep, put their head in a frame that holds the skull very rigidly, and then you take a CT scan of their head while they are sleep, put that frame on and then merge, uh, the M, R I and A C T and software.

You have A A plan based on the M R I, where you can see these nuclear deep in the brain. You can see the on C. T, but if you trust the merging of the two images, then you indirectly know on the city where that is, and therefore indirectly know where, in reference to the titanium frame screw to their head, those targets are.

And so this is sixty technology to manually compute trajectories, given the entry point and target a and dialed some goofy looking titanium um actuators um with manually manual actuators with little tick Marks on them. The modern version of that is a robot, uh you know, just like A A little cooker ARM. You might see a building cars at the tesla factory.

Uh, this small robot ARM can show you the trajectory or that you intended from the P O M R I, and establish a very rigid holder through which you can drill a small hole in the skull and pass a small rigid wire deepen in that area, the brain that's hello, and put your electoral through that holo wire and then remove all of that except the elector. See, you know, put the elector very, very precisely placed far from the school surface. Now that standard technology that's already, you know, been out the world for for a while nurlan right now is entirely on cortical targets, surface targets uh, because there's no trivial way to get, say, hundreds of wires deep inside the brain without doing a lot of damage.

So your question, what do you see? What I see A M R. I on the screen, I can't see everything that that D B S. Elector is passing through on its way to that deep target.

And so it's accepted with this approach that there's gonna be about one and one hundred patients who have A A bleed somewhere in the brain ah as a result of passing that wire blindly into the the deep part of the brain. That's not an acceptable safety profile for neural ink. We uh start from the position that we want this to be dramatically maybe two or three orders of magazine de safer than that. Uh safe enough really that you know you are I without a profound medical problem might on our lunch break some day say, sure, i'll get that been meaning to upgrade to the latest version and so that the safety constraint given that are high. And so we haven't, uh, settled on a final solution for arbitrarily approaching deep targets in the brain.

It's interesting is like you have to avoid blood us somehow you have to maybe there's creative ways to do the same thing like mapping out high resolution geometry blood vessels and then you can go in blind. But I how do you map about that in with the six super table that there's a lot interesting chAllenges there, right? Yeah, but there's a lot to do on the surface.

exactly. So we've got vision on the surface. Um you know we actually have made a huge amount progress showing uh electrodes into the spinal cord as a potential work around for a spinal court injury that would allow a brain mounted in plant to translate motor intentions to a buy mounted in plant that can effect muscle contractions in previously paralyzed arms and .

that's just incredible. So like the effort there is to try to bridge the brain to the spinal cord to the proph oner nervous. So uh how hard is that to do?

We have that working in ah in very crude forms and .

animals that's amazing. A similar know we he's able to digitally move the cursor here you're doing a the same kind of communication but with the actual the factors that you have yeah that's fast. yes.

So we have unethical zed animals doing grasp and moving moving their legs and and walking pattern. Again, early days, but the future is bright for this kind of thing and people with paralysis, I should look forward to that right future. They're to have options.

Yeah, there's a lot of sort of intermediate or extra options where you take like an optimist robot like the uh the ARM to B O to control the ARM yes the the the fingers the hand of the ARM is a prostatic .

are getting Better .

to yeah so that that goes hand in hand. Although I didn't quite understand til thinking about a deeper and do more research about neural link, how much you can do on the digital side. So there's digital telepathy, I didn't quite understand that you can really map the intention, as you described in the hand knob area, then you can map the intention.

Just imagine IT think about that intention can be mapped actual action in the digital world. And now more more so much can be done than in the digital world that IT IT IT can reconnect you to to the outside world that can allow you to have freedom, have independence, if you are a quite a plagiary. Yeah, that's really powerful. I can go really far with that.

Yeah our first participant is is he's incredible. He's breaking world records .

left and right and he's having phone with IT. It's great. Just going back to the surgery your whole journey.

You mentioned me to find your surgery on monday. You like doing surgery all the time. Yeah, maybe the ridiculous question, what does IT take to get good at surgery?

Practice repetitions? You just say with anything else, you know, there's a million ways of people saying the same thing and selling books saying that. But you call IT ten thousand hours. You call IT spend some chunk of your life, some percentage of your life focusing on the succession about getting Better at IT um repetitions, uh humility, recognizing that you aren't perfect at any stage along the way, uh, recognizing you've got improvements to make in your technique, being open to feedback and coaching from people with a different perspective on how to do IT um and then um just the constant will to do Better than fortunately, you know if you're not associated path, I think your patients bring that with them to the office visits every day. They force you to want to a do Better all the time.

Yeah to step up. I mean, the real human being, a real human being that you can help yeah. So every surgery, even if is the same exact surgery, is there a lot of variability between that surgery and a different person?

Yeah, a fair bit. I mean, a good example for us is that the angle of the skull relative to the Normal play, know of the body access of the skull overhand nob is pretty wide variation. I mean, some people have really flat schools, and some people have really deeply angled skills over that area.

And that has you know consequences for how their head can be fixed in in a in sort of the frame that we use um and how the robot has to approach the skull. And yeah people's people's bodies are built as differently as in other people who see walking down the street as as much variability and body shape and size as you see there we see in brain anatomy and kull anatomy. Um there are some people who we've had to kind of exclude from our trial for having skills that are too thick or too then or scalp that's too thick or too then. Um I think you know we have like the middle ninety seven percent or so uh of people, but you can't account for all human anatomy variability how .

much like machine and messes there because you know taking biology classes, the dagger grounds are always really clean and crisp. Neuroscience, the pictures of neuron there was really nicer. But whenever I look at pictures of like real brains though I I don't know was going on yeah so how much are biological systems in reality? Like how hard is IT to figure out was going on?

Not too bad at once. You've really get used to this. You know that's where experience and and scale and education really come into play. As if he starred a thousand brains, IT becomes easier to kind of mentally peel back the, say, for instance, blood vessels that are obscuring the salsa and g on the wrinkle pattern of the surface of the brain.

Occasionally, when you, when you first starting to do this and you open the skull, IT doesn't match what you thought you were going to see based on the mr. eye. And with more experience, you you ve learned a kind of peel back that layer blood vessels and see the underlying pattern of wrinkles in the brain and use that as a landmark for where you are.

The wrinkles are a landmark.

So like, yeah, so I was describing hand, not earlier. That's a pattern of the wrinkles in the brain, and sort of this sort greek letter omega shaped area of the brain so you could .

recognize the hand nob arria. Like, if I assure you, thousand brains give you like one minute with each you be like gap. That's nature.

And so there is some uniqueness to that area of the brain. Like in terms of the geometry, the topology of the thing. Yeah, where is IT about in it's so you .

have this strip of brain running down the top that called the primary motor area. And sure, you've seen this picture of the humongous later, over the surface of the brain, we were little guy with huge lips and giant hands. A that guy sort of lays with his legs up at the top of the brain and face ARM, uh, areas farther down, and and then some kind of mouth, lip, tongue areas, uh, farther down.

And so the hand is right in there. And then the areas that controls speech, at least on the on the left side of the brain. And most people are are just below that.

And so are any muscle that you voluntarily move in your body. The vast majority that references that strip where those intentions come from. That's just a brain. And the the wrinkle, uh, forehand ob is right in the middle. That and vision .

is back also on close to the surface vision's .

little deeper. And so you know this kids, to your question about how deep can you get um to do vision we can adjust to the surface of the brain. We have to be able to go in, uh not not as deep as we have to go for D B. S, but maybe a centimeter deeper than we're used to forehand insertions. Uh and so that you know, work in progress, that uh, a new set of chAllenges overcome.

By the way, you mentioned the utah right and I just saw picture that and that thing looks terrifying because it's it's because it's rigid. And if you look at the threats, they are flexible. What you say that's interesting to you about the flexible that kind of approach of the the flexible threads to deliver the electrodes next to the neons?

yeah. I mean, the goal there comes from experience. I mean, we stand on the shoulders of people that made you tories and and used you tories for decades before we have even came along. Nerlinger rose partly. This approach to technology rose out of a need recognized after uti.

I would fail routinely because the original electrodes, those Spikes that are literally hammer using air hammer into the brain, the Spikes generated bad mune response that encapsulates the the electorate Spikes in, uh, scar tissue essentially. And so one of the projects that was being worked on in in the Anderson lab c caltech, when I got there, was to see if you could use chemo therapy to prevent the formation of scars. Like, you know, things are pretty bad when you're gaming a bit of nails into the brain and then 确定 that with chemotherapy to try to prevent scar tissues， like, you know, maybe we've gotten off track here, guys.

Maybe there's a fundamental redesign necessary and so neural ink approach of using highly flexible, tiny electrodes avoids a lot of the bleeding, avoids a lot of the immune sense that ends up happening when rigid electrodes are pounded into the brain. And so what we see is our electorate longevity and functionality, uh and and the health of the brain tissue immediately surrounding the electrode, uh, is excEllent. I mean, that goes on for for years now in our animal models.

What do most people not understand about the biology of the brain in the vascular is sure that's really interesting. I think the .

most interesting may be under appreciated fact. Yeah, is that IT really does control almost everything. I mean, I don't know you out of a blue example, imagine you you want a lever on fertility.

You want to be able to turn fertility on enough I there are legitimate targets in the brain itself to modulate fertility, say um blood pressure. You wants to moderate blood pressure. There are legitimate targets in the brain for doing that um things that aren't immediately obvious as brain problems are potentially solvable in the brain. Um and so I think it's an under explored area for primary treatments of a of all the things that bother people.

It's a really fascinating way to look at IT, like there's a lot of conditions we might think have nothing do with the brain, but they might just be symptoms of something that actually started in the brain. The actual source of the problem, the primary source .

is something yeah not not always. I mean, you know kidney disease is real. But um there are levels you can pull in the brain that affect all of the all of these systems.

this nobs on off switches in jobs in the brain from which this all or original ates ah would you have a new link chip in planted .

in your brain？ Yeah I think .

youth case .

right now is use a mouse, right? I can already do that. And so there's no value proposition uh, on safety grounds alone sure. I'll do IT tomorrow.

say the use case of mouse. Is after like researching all this, and part of this is just watching now and have so much fun if you can get that bits per second look really high with a mouse, like being able to interact here. If you think about the way the and smart phone we wipe, that's transformational.

Yeah holding interact with the thing is set of you don't realize the able to touch your phone and to a score with your finger that's like that changed everything that the people were sure you need a keyboard to type in the that, uh, there's a lot of H C, I aspects to that, that changed how interact with computers. So there could be a certain rate of speed with a mouse that will change everything. Yes, like you might be able to just click on a screen extremely fast. And that.

if IT I can.

must have in a neural link for much more rapid interaction with digital devices.

Yeah, I think recording speech intentions from the brain might might change things as well. The value proposition for the average person a keyboard is a pretty clunk y human inter face or requires a lot of training. Its you highly variable in the maximum performance that the average person can a can achieve. I think taking that out of the equation and just having a natural in a word, to computer interface might change things for a lot of people.

That would be hilarious if that is the reason people do IT, even if you have speech to taxes. Extremely accurate, equally isn't. But I say gotten super accurate. IT be hilarious. Y people went for neural link. Just see you avoid the embarrassing aspect of speaking like, looking like a dog back, speaking to your phone in public, which is a real like that's a real constraint .

yeah I mean with a bone conducting case, uh that can be an invisible headphone, say and the ability to think words into software and have IT respond to you um you know that starts to sound sort of like embedded super intelligence you know, if you can silently ask for the wikipedia article on any subject and have IT read to you without any observable change happen in the outside world. You know, for one thing, standardize testing is obsolete yeah .

if it's done well in the oxide de IT could change. I don't have transformed society, but IT really uh can create a kind of shift and way we interact with digital devices in the way that smart founded. Now I would just having to look into the safety of everything involved, though I would totally try IT.

So IT doesn't have to go to some like incredible thing. We have a connects your vision or to some like you connect all of your brain that could be like just connecting to the hand nob. Uh, you might have a lot of interesting interaction, human computer interaction possibilities. yeah.

That's interesting. yeah. And the technology on the academic side is progressing a light speed here. I think there is a really amazing paper out of U C. Davis, uh surveys to whisk lab that basically made a initial solve of speech decode. There was something like a hundred and twenty five thousand words that they were getting with, you know, very high accuracy.

which is so you you're just thinking the word yeah thinking the word nearly to get IT yes. Or by like you have to have the intention of speaking IT right to like, do that inner voice now is so amazing to me that you can do the intention, the signal mapping I have to do, just imagine us of doing IT.

And if if you get the feedback that I actually worked, you can get really good at that like your brain will first love, just you develop like any other skill yeah like touch typing, you develop in the same kind of way that is to me. Is this really fascinating you to be able to even to play with that, honest, like I will get a really just to be able to play with that, just to play with the capacity, the capability of my mind to learn the skill. It's like learning the skill of typing and learning the skill of moving a mouse. It's another skill of moving the mouse, not with my physical body, but with my mind.

I can't wait to see what people do with that. I feel like, or cave man right now were like banging rocks with a stick and thinking that we're making music at some point when these are more widespread. This can be the equivalent of of a piano that, you know, someone, someone can make art with their brain in a way that we didn't even anticipate. I'm looking ford to IT go IT to .

like a teenager, like any time I think i'm good at something I I was good like I don't know, even even even with the the debit per second and playing a video game, you realize you go to A T, you give in your link to a teenager. Just a large number of them, kind of they get good at stop. They're gonna get like hundreds of um this per second. We even just for the current technology.

probably probably because it's .

also addicting how all like the the the number go up aspect a bit of like improving and training because IT is is almost like a skill and is the software on the other end that adapts to you and especially the adopting procedure algorithm becomes Better, Better, Better. You like learning together. Yeah.

we're scratching the surface on that right now. There's so much more to do.

So now complete other side of IT. You have an alpha d chip up implanted in you ah was so I here yes. So this is the passle thing is the possible device that you use for unlocking like a safer top secrets. What is what for? What's the story behind IT?

I'm not the first one. There's there's a little community of weird w biohackers that um have done this stuff. And I think one of the early youth cases was storing you know private crypto wallet keys and and whatever. I doubled in that a bit and and had some fun with IT get some big going implanted .

in your bodies and where you can tell where yeah yeah .

actually yeah uh you know the modern equivalent of finding change in the so focus after I I put some or clipped on there that I thought was worthless and forgot about IT for a few years, went back and found that some community of people loved IT and had propped up the value of IT. And so IT had gone up fifty fold. So there was a lot of change in those questions.

It's a literis, but the primary youth cases, mostly as a, as a tech demonstrator. U IT. Has my business card on, you can scan that. And uh, by touching IT to your phone that opens the front door of my house. You know, whatever simple stuff with a .

cool step is a cool leap to implant something in your body. I mean, IT has perhaps that is a similar leap to a neural link, because for a lot of people, that kind of notion of putting something inside your body, something electronic inside of biological system.

is a big leap. Yeah, we a kind and the barrier of our skin, we're completely fine with new replacements hyp replacements you know uh dental implants um but um you know there's a missed ism still around the invariable barrier that the score represents, and I think that needs to be treated like any other pragmatic barrier. The question isn't how how incredible is that? Open the score. The question is, you know, what benefit can we provide?

So from all the surges down for everything, understand the brain, how much does neuroplasticity come to play? How adapt poles, the brain, for example, just even in the case of healing from surgery or adapting to the poll surgery situation.

that is sad for me and, uh, other people. My demographic is that, you know, the plasticity decreases with age. Healing decreases with age. I have too much gray hair to to be optimistic about that.

There are theoretical ways to increase plasticity using electrical stimulation nothing that is you totally proven out as a robust enough mechanism to offer widely to people. But um yeah I think I think there is cause for optimism that we might find something useful in terms of say, and implanted electorate that improves learning. Um certainly there's been really amazing work recently from a niclas shift.

JoNathan Baker, you know others uh who have A A cohort of patients with moderate traumatic brain injury who have had electrodes placed in the deep nucleus in the brain of the central media nucleus or just near the central media nucleus, and when they apply small amount of electricity to that part of the brain, it's almost like electronic caine. They're able to improve people's attention and focus. They're able to improve how well people can perform a task.

I think in one case, someone who was unable to work after the device was turned on, they were able to get a job. Uh and that sort of, you know one of the holy grills for me with neural link and other technologies like this is from a purely util italian standpoint. Um can we can we make people able to take care of themselves and their families economically again? Can we make IT? So someone who's fully dependent and even maybe requires a lot of caregiver resources, can we put them in a position to be fully independent, taking care themselves, giving back to their communities? Um I think I think that's a very compelling proposition and what motivates a lot of what I do and what a lot of the people at newlin k are working for.

It's just a cool possibility that if you put a new link there that the brain adapts like the other part of the brain adopts too no IT integrates the capacity of the brain to do that really interesting, probably unknown degree to which you can do that. But you're now connecting an external thing to IT especially once is doing uh stimulation like the the biological brain and the ah the electronic brain outside of IT working together they could the possibilities they're really interesting are still unknown but interesting IT feels like the brain is really good at adapting to whatever yeah but of course this is a system that by itself is already uh like everything serves the purpose and so you don't miss with that too much yeah it's like.

you know eliminating a species from from an ecology you know you don't know what the delicate interconnections independency are um the brain is certainly a delicate complex beast and .

we don't know uh you know every .

potential downstream consequence of a single change that we .

make d see yourself doing so mention p one surgeons of p two, p three, four, p five. Just well, more, more, more humans. I think you .

know it's a certain kind of brittleness know a failure on the company's side. If we need me to do all the surgeries, I think something that I would very much like to work towards is a process that is so simple and so robust on the surgery side that literally anyone could do IT. Um we we want to get away from requiring intense expertise or intense experience to have this successful ly done and make IT as as simple and translatable as possible.

I I mean, I would love IT if every neurons on the planet had no problem doing this. Um I think we're probably far from a regulatory environment that would allow uh, people that aren't know a surgeons to do this. But uh, not impossible.

I all sign ough for that. Did you ever enter promotional es the robot or one like to give you her name to see you as like a friend, as like working together with you?

I've been to a certain degree, or an enemy who 对。 To a certain degree is it's complex relationship. All the .

good relationships are.

It's funny when in the middle of the surgery, there's a part of IT where I stand basically shoulder to shoulder with the robot. And so you if you're in the room reading the body language, you know it's my brother in arms and were working together on the same problem yeah, i'm not threatened by IT.

Keep telling yourself that is how have all the surgeries that you've done over the years, the people you've helped in that the stakes, the high stakes that you ve mentioned, how how's IT change your understanding of life and death?

Yeah, i'm. You know, IT gives you a very visceral sense, and this may be sound try, but IT gives you a very visceral sense that death is inevitable. You know, on one hand, you know you you are as a newer surgeon, you deeply involved in these like just hard to find them tragedies um you know Young parents dying living you know, a four year old mind say.

And and on the other hand, you know IT takes the sting out of IT a bit because you see how just my, namely ly universal death is there is zero chance that i'm going to avoid IT uh, I know you know techno optimists right now and longer gev ity buffs right now would disagree on that zero point zero zero zero percent estimated, but I don't see any chance that our generation is going to avoid entreprise, a powerful force. And we are very ornate, delicate, brittle, DNA machines that aren't up to the cosmic ray bombardment that were subjected to. So on the one hand, every human that has ever lived, died or will die.

On the other hand, it's one of the hardest things to imagine inflicting on anyone that you love is, is having them gone. I mean, i'm sure you've had friends that aren't living anymore and it's it's hardly even think about them. And so um I wish I had you know arrived .

at the .

point of nevada where you know death doesn't have a thing. I'm not worried about IT, but I can at least say that i'm comfortable with the certainty of IT if not having found out how to take the the tragedy out of IT when I think about you know my kids uh either not having me or or me not having them or my wife maybe .

i've come to accept the intellectual certainty of IT. But IT may be the pain that come to lose in the people you love. But I think i've come to understand the existential aspect of IT like that. This is gonna a end and I don't mean like in some tight way.

I me like it's certainly feels like it's not going to end like you live life like it's not going to end, right and the fact that this light that's shining, this consciousness is is is going to um no longer be one moment. Maybe today is like a IT feels me when I really am able to at all that in with ernest becker terror that is a real fear. I think people aren't always honest with how terrifying IT is.

yeah. I think the more you are able to really think through IT, the more terrifying IT is is is not such a simple thing. I was the way life isn't. If you really can mode that in, it's hard.

But I think that's why they still exit IT because it's like hopes you get your shit together and be like the the moment every single moment your alive is just beautiful, yes, and it's terrifying that is going to end. This is the like, almost like you're shivering in the cold, a child helpless, this kind of feeling. And then that makes you, when you have warmth, when you have the safety, when you have the love, really appreciate IT.

I feel like sometimes in your position when you mentioned armour, just to see that you might make you not be able to see that the the findings of life because if you kept looking at that, you might break you. So it's it's good to know the year kind of still struggling with that. There's the new surgeon and there is a human yeah and the human is still able to struggle, then feel the the fear of that in the pain of that .

yeah you know IT definitely makes you ask the question of how long, how many time, how many of these can you see and and not say, I can't do this anymore. But I mean, you said at well, I think that gives an opportunity to just appreciate that you're alive today. And you know, i've got i've got three kids and an amazing wife and i'm really happy.

Things are good. I get to help on a project that I think matters. I think that moves us forward. I'm a very lucky person.

It's the early steps of a potentially gigantic leap for humanity. It's a really interesting one. And it's cool as like you, you read about all the stuff in history, words like the early days been reading before, gone to the amazon now read about explores, uh they will go and explore even the amazon and jungle for the first time.

Just doesn't early steps, yes, or early steps in the space, early steps in any discipline in in physics and mathematics and and cook, as this is like the underground scale, easy. The early steps into dulling, deep into the human brain. So not just observing the brain, but be able to interact with the human brain. It's going to help a lot of people, but they also, may I help us understand what that has gone on in there?

Yeah I think ultimately, we want to give people more levers that they can pull, right? Like you wanna give people options if you can give someone a dial that they can turn on how happy they are. I think that makes people really uncomfortable.

But um now talk about major depressive disorder, talk about people that are committing suicide at an alarming rate in this country. and. Try to justify that quiz.

Ss, in those, in that light of you, you can give people a nob to take away suicide ideation, suicide intention. I would, I would give them that. Nab, I don't know how you justify not doing that.

They can think about, like all the suffer is gone on the world, like every single human being that suffering right now that be a glowing red dot. The more suffer, the more is glowing. And you you see the map of human suffering and any technology that allows you to dimmed that light of suffering uh and a grand scale.

Is this pretty exciting? There's a lot of people suffering and most of them suffering quietly. And we turn our uh we we look away too often uh and we we should remember those are suffering because IT once again, most of them are suffering quietly well.

And you know, on a grander scale, the fabric of society. You know, people have a lot of complaints about how our social fabric is working or not working, how our politics is working or not working.

Uh, those things are made of our chemistry to in in aggregate, right, like our politics is composed of individuals with human brains and the way that works or doesn't work is potentially tunable uh, in the sense that I don't know, say, remove our addictive behaviors or tune our addictive behaviors for social media or addiction to outrage our addiction to sharing the most angry political tweet we can find, I don't think that leads to a functional society. And if if you had options for people to moderate that my adaptive behaviour, there could be huge benefits to society. Maybe we could all work together a little more harmoniously toward useful ends.

There's a sweet pot like you mention. You don't want to completely remove all the dark side of human nature because those kind of are somehow necessary to make the whole thing work.

But there's a sweet spot yeah, I agree to suffer a little, just not so much that you lose hope .

yeah when you all the surgeons you've done housing consciousness and there ever was there like a glowing light. You know, I have .

this sense that I never found IT never removed IT. You know, like like a demented orr Harry potter. I have this sense that consciousness is a lot less magical than than our instincts.

Wana claim IT is um IT IT seems to me like a useful analogue for thinking about what conscious ness is in the brain. Is that we we have a really good intuitive understanding of what that means to say to touch your skin and know what's being touched. Um I think consciousness is just that level of sensory mapping applied to the the thought process is in the brain itself.

So what i'm saying is consciousness is the sensation of some part of your brain being active. So you you feel IT working. You feel the party, your brain that thinks of read things or wind creatures or the taste of coffee. You feel those parts of your brain being active, the way that I am feeling, my poem being touched, right? And that sensory system that feels the brain working is consciousness .

so brilliant, it's the same way. Is the sensation of touching when you're touching a thing? Consciousness is the sensation of you feeling your brain, working, your brain thinking, your brain perceiving which .

isn't which isn't like a warping of space timer or some quantum field effect, right it's nothing magical. People always want to describe to consciousness something truly different um and there's this awesome long history of people looking at whatever the latest discovery and physics is to explain consciousness because it's the most magical, the most out there thing that you can think of and and people always you know want to do that with consciousness. I don't think that's necessary. It's just you know a very useful and gratifying way of feeling your brain work and as we said.

as one hook of a brain, yeah everything we see around as everything we love, everything is beautiful, came from brains like these.

It's all electrical activity happen inside your skull.

And I for one am a grateful those people like you there, uh, expLoring all the ways that IT works and all the ways that can be made Better. Thank you so much for talking. Thanks for listening to this conversation with Matthew mcduck.

U and now, dear friends, here's bliss chapman rain interface software lead at new link. You told me that you've met hundreds of people, will spin a cord injuries or with us, and that you're motivation for helping in your link is grounded. Wanted to help them. Can you describe this motivation?

yeah. First, I just to thank you to all the people i've gone a chance to speak with for sharing their stories with me. I don't think there's any world dely in which I can share the stories as powerful ways they can. But just, I think, to summarize a very high level, what I hear over and over again is that people with, uh, L, S, or severe according tree in the place where they place can't move physically anymore, really, at the end of the day, are looking for independence. And that can mean different things for different people, for some folks.

That can mean the really just to build, communicate again independently without needing to wear something on their face, without needing a career real, real to put something in their mouth, for some folks, can mean independence will able to work again to be able to navigate a computer digitally efficiently enough to be able to get a job, to able to support himself, to be able to move out and alto, to be able to support themselves after their family. Maybe isn't there anymore to take care of them, and for some focus as simple as just being hold to respond to the kid in time before they, you know, run a way or get interested in something else. And these are deeply personal and very human problems.

And what strikes me again and again when talking with these folks, is that this is actually an engineering problem. This is a problem that with the right resources, with the right team, we can make a lot of progress on. And at the end of the day, I think that's a deeply inspiring message and something that excited to get up everyday.

So as both an engineering problem in terms of A, B, C, I, for example, that can give them capabilities where they can interact with the world. But also on the other side is an engineering problem for the rest of the world to make IT more accessible for people living with quarter politica.

Yeah and I see all take a broad view lands on this first. Second, I think i'm very in favour of anyone working in this problem space. So beyond C, I am am happy and excited and willing to support.

Anyway, I can folks working on tracking systems, working on you know, speech to tech systems, working on head trackers or mouse sticks or quarter sticks. And I have met many engineers and folks in community that do exactly those things. And I think for the people were trying to help IT.

Doesn't matter what the complexity the solution is, as long as the problem is solved. And I on emphasize that there can be many solutions out there that can help with these problems. And a bci is one of a collection of such solutions.

So B, C, I in particular, I think, offer several advantages. Ges here. And I think the folks that recognizes the the media, usually the people who have, want to create or some reign policies, usually you don't have to explain to them why this might be something that could be helpful.

It's usually pretty self evident. But for the rest of us, folks don't live with severe sponsor or injury or who don't know somebody with a us. It's not often obvious why you would want a Brant plant to be able to connect, navigate a computer, and it's surprisingly new.

One thing to degree that i've learned a huge amount, just working with no land in the first nearly in clinical trial and understanding from him, in his words, why this devices in powerful for him. And it's a new old topic, IT can be the case that even if you can achieve the same thing, for example, with a mouth stick when navigating in computer, he doesn't have access to a masc every single of the day. Young, he has access when some was available, to put IT in front of him. And so obese, I can really offer a level of of independence and the autonomy that if IT wasn't literally physically part of your body, be hard to achieve any other way.

So there is a lot of facing aspects to what IT takes to get no one to be able to control a curse Green with his mind. You text to me something that I just love, he said. I was part of the team that interviewed and selected p one.

I was in the Operating room during the first human surgeon, moderating life signals coming out of the brain. I work with the user basically every day to develop new U. S. Paradise coding strategies.

And I was part of the team that figured out how to recover useful bci to new world record levels on the signal quality degraded will talk about, I think, every aspect to that. But just zoom ing out. What was I like to be part of that part, part of that team and part of that historic, I would say, historic first.

Yeah, I think for me, this is something I had an excited about for close to ten years now. And so to be able to be even just some small part of making IT a reality is extremely exciting. A couple may be special moments during that whole process that i'll never really, truly forget. One of them is in turning the actual surgery.

Um you know at that point in time I I know no and quite well I know his family and so I think the the initial reaction when uh no is rolling in the Operation is just uh oh shit kind of reaction but at that point muslim memory kicks and and you should have go to a you actual body just do that of the of the talking. And I have a lucky job in that particular procedure to just be in charge of Mandatory ing the implant so my job is is set there to look at the singles coming off the implant, to look at the life brain data streaming off. The device has to add being inserted to the brain, and just to basically observe a mixture that nothing is going you know wrong, or that this no problem, flags are falling, conditions that we need to go investigate, our poses surgit to debug.

And because I had that a respective review of the surgery, I had a slightly remove perspective. And I think most folks in the room, I got, sit there, anything to myself, wow. You know, that brain is moving a lot.

Now, when you really look inside, look to me that we threads and, you know, wanting that most people don't realizes, the brain moves. The brain moves a lot when you breathe, when you, when you are hard beats. And you can see IT visibly. So you know, that's something that I think was a surprise to me and very, very exciting to to see someone's brain who you physically know. And you've talked with that length actually posting in moving inside their school.

and they use that brain to talk to you previously. And I was, right, right. They are moving.

Uh, actually I didn't realize that in terms of the threat running. So the the neural and implant is active during surgery. So in one thread at the time, you're able to start seeing signal. So that's part of the way you test that the thing is working.

yes. So actually in the in the Operator right after we uh finished the all the insertions, I started collecting what's called broadband data. So broadband is uh basically the most raw form of single. You can clock from a nearly electrode is essentially a measured of uh local field potential or the yeah the volt essentially measured by the elections and h we have a certain mode in our international licence that allows us to visualize where detective bikes are. So IT visualize this sort of where uh in the broad band is very, very reform of the data, a neuron is actually spiking. And so one of the the this moment that i'll never forget his part, this whole clinical trial is seeing live in the Operating room while he's still under nazia beautiful Spikes being shown in the application to streaming live to devise some holding in my hand.

So this is no signal process. The raw data and in the signals process is on top of that. You're seeing the Spike detected, right? And that's the ux too, because, yes, that looks beautiful as well. During that predation.

there is actually about a camera men in the room, so they also were curious and want to see. There are several other surgeons in the room who are all just excited to see robots taking their job. And there are crowded around the small of iphone, watching this life, brining a stream of his spring.

What was that like? Seeing the robot to some of the surgery. So the computer vision aspect were detects all the all the posts that avoid the the blood vessels and then of the human supervision and actually doing the really high precision, uh, connection of the threads to the brain. Yeah.

had a good question. My answers is, gonna be pretty lame here, but I was boring. Yeah, i've seen IT, uh, so many times.

Yeah, that's exactly you want, if you wanted to be boring. Yeah, because i've seen that so many times. I seen the. The robot do this surgery literally hundreds times, and so was just one more time.

Yeah, all the practice surges of proxies. And just another day. no. So what about when I woke up? Well, did you remember a moment where a he was able to move the curse, or not move the cursor, but get signal from the brain, such that I was able to show that there's a connection?

Yeah, yeah. So we are quite excited to move as quick this weekend. And no one was really, really excited to get started. He wanted to get started actually the day of surgery but um we we waited till the next morning very patiently so long night um and the next morning in the I C U where he was uh recovering, he a wanted to get started and actually start to understand what I have seen no we can measure from his brain.

And folks who are not familiar with um the earth system we implant the neural ink system are the earthling implant in the motor cortex. So the motor cortex is responsible for representing things like motor intent. Um sorry if you imagine closing and opening your hand, that kind of single representation would be present in the motor.

Tex, if you imagine moving your arms back and forth or wigless a pinky, this sort of signal can be present in water critics. So one of the ways we start to sort of map out what kind of signal do we actually have access to in any particular individuals brain is through the task called body mapping and buying up this, where you essentially present a visual to the user, you say, hey, imagine doing this. And the visual is, you know a three hand opening, closing or index finger module up and down.

And h, you ask the user to imagine that. And obviously, you can't see them do this because they're paralyzed. So you can see that actually move their ARM.

But while they do this task, you can record in your activity, and you can basically offline model and check. Can I predict or can I detect the modulation correspondence with those different actions? And so we did that task.

Can we realize, hey, there's actually some modulation associated with some of his hand motion, which was the first indication that OK, we can potentially use that modulation to do useful things in the world. For example, control a computer cursor. And he started playing with that.

You know, the first time we showed him him and we actually just took the same life view of his practice ity and put IT in front of him and we said, hey, you tell us what's going on. You know, we're not you you're able to imagine different things and we know that it's modulating some of these neurons. So you've figure out for us what that is actually representing.

And so he played with IT for a bit. He was, the eyes don't quite get IT yet. He played for a bit longer.

And he said, oh, when I move this finger, I see this particular dance started fy more. And I said to prove, do again. And so I said, okay, three, two, one, boom.

And the minute he move, you can see like instantaneously, this neuron is firing single neuron. I can told the exact channel number of you, you're interested, stuck in my brain now forever. But that single, uh, channel firing was a beautiful indication that IT was behaved very modulated neutral activity that could then be used for downs room tasks like decoding computer.

And we say single channel. Is that associated a single electron?

Yes, a channel electoral interchange table .

and there's a one thousand twenty four of those one .

thousand twenty four, yes.

I think about that. That works. That really, when I was learning about all this, and like loading IT in IT, was just blowing my mind that the intention you can visualize yourself moving the finger that can turn until signal.

And the fact that you can then skip that step and visualize the course are moving, or have the intention of the cursor moving in that leading to a signal that can then be used to move the curse. There are so many exciting things there to learn about the brain, about the way the brain works. The very fact of their existing signal that can be used is really powerful.

But IT feels like that's just the beginning of figuring out how that in be used really, really effectively. Now I should also just there's so many fascinating details here, but you mentioned the body mapping step, at least in the version I saw that no one was showing off. There's like a super nice interface, graphical interface. But like, I just felt like I was like in the future because like you, I guess I visualizes you move in the hand and there's a very like like a sexy polish interview. Hello, I don't know if there's a voice component, but I just felt like like when you wake up in a really nice video game and this is the toil, the beginnings of that video ah this you are supposed to do.

it's cold. No, I mean, the future should feel like the future.

but it's not easy to pull that off. I mean, need to be simple.

but not too simple. Yeah, I think the U X design compound here is uh, under rated for B, C, I, H development in general. There a whole interaction effect between the ways in which you visualize uh, and instruction to the user and the kinds of single you can get back.

And that quality of sort of your behavior aligned to the is a function of how good you are expressing to the user, what you want them to do. And so we have we spend a lot of time thinking about the ux, a of how we build our applications, ions of how the decoder actually functions, the control surface that provides the user. All these little details matter alone.

So maybe would be nice again to a little bit more detail of what the signal looks like and what the decoding looks like. So there's a and one implant that has like to mentioned a one thousand twenty four electrodes s and that's collecting raw data road. What is that single lake and uh, what are the different steps along the way before is transmitted, ted? And what is transmitted?

All that kind of stuff? Yeah, yeah, this is going to be a fun one. So, uh so, uh, maybe before diving into what we do is worth understanding what we're trying to measure because, uh, that dictate tes all the requirements for system that we build.

And what we're trying to measure is really individual neons producing action potentials and action potential is you can think of IT like a little electrical impulse that you can uh detect if you're close enough. And by being close enough, I mean, like within, let's say, a hundred microns l and hundred mrs, is very, very tiny, distant. And so the number of neurons that you are going to pick up with any given a lot trade, is IT just a small radius around that electra.

And the other thing worth understanding about the underlying biology here is that when neurons producing potential, the way of that action potential is about one thousand. So from the start, the Spike to the end of the bike that whole with of that um of characteristic feature of a non firing is one millisecond wide. And if you wanted attack that an individual Spike is occurring or not, you need to sample that signal or sample the local fuel potential nearby neon much more frequently than once a mili can.

You need to sample many, many times for miles I can, to build a tech that this is actually the characteristic way form of a neuron producing an action potential. And so we sample across all thousand twenty four last trades about twenty thousand times a second. Twenty thousand times a second means already given one more second window with about twenty samples that tell us what that exact shape of that actually potential looks like and one, two sort of sample ed, a super high rate uh, underlying logical field nearby.

Uh, the cells we can process no into just where do we to touch Spike or where do we not? So miners in a one or zero do we to touch back in this one mile, second or not. And we do that because the actual information character Carrying a sort of subspace of neural activity is just winner Spike occurred.

Essentially everything that we care about for decoding can be captured, represented in the frequency care eristic s of Spike trains, meaning how often are Spikes firing in any given window of time. And so that allows us to do sort of a crazy amount of compression from this very rich identity, uh, you know, signal to something that much, much more sparse and compressible that can be sent out over a wireless radio michabo to communication. For example.

quick tensions. Here you mentioned electrode neuron. There is a local neighbor od of neurons nearby. How difficult is that too? Like isolate from where the spite came from yeah so there's a whole .

field of the academic neuroscience work on exactly this problem of basic given a single electrode or given a set of electro des measuring uh set of neurons, how can you sort Spike sort which Spike are coming from what on? And this is a promise pursuing academic work because you care about IT for understanding what's going on in the underlying through, uh, newer sites of the of the brain.

If you care about understanding how the brain are presenting information, how that evolving through time, then that a very, very important question to to understand. For a serve the engineering side of things, at least at the current scale, if the number of neurons per electro de is relatively ly small, you can get away with basically more in that problem completely. You can think of IT, like sort of a random projection of neurons to electrons.

And there maybe in some cases more than wonder on perlite ude. But if that number is small enough, those cynical can be thought of as, uh, sort of a union of the two. And for many applications, are a totally reasonable trade off to make and can simplify the process a lot.

And as you serve scale out channel count, the uh relevance of distinguishing individual neuron becomes less important because you have more overall signal and you can start to reliance of correlations of cover structure in the data to help understand win that channels firing, what is that? What is that actually represent? Because you know that when that channels firing in concert with these other fifty channels, that means move left. But when that same channels firing with concert with these others and channels have to move right.

okay, we have to do this kind of speed detection on board, and you have to do that super efficient, so fast and not use too much power because you don't want to be generate too much heat. So to be a super simple signal processing step, yeah is there some wisdom you can share about what IT takes to overcome that chAllenge?

Ah so we've tried many different versions of. Basically turning this raw signal into, uh, through a future that you might want to send off the device. And i'll say that I don't think we're at the final step of this process.

This is a long journey. We have something that works clearly today, but there can be many approaches that we find in the future that are much Better than what we do right now. So some versions of what we do right now, and there's a lot of academic, is to these ideas.

So I don't want to claim that these are original, nearly idea as anything like done. But uh, one of these ideas is basically to build up sort of like a convolution filter. Ams, if you will, that slide across the single and looks for a certain template to be matched.

Temple consists of of how deep the Spike modulates, how much a recovers and what the duration window of time is that the whole process takes. And if you can see the single that temple is matched with them certain bounds, then you can say, okay, that a Spike. One reason that approaches super convenient is that you can actually implement that extremely efficient in hardware, which means that you can run IT, uh, low power across a thousand twenty four channels.

The ones are another approach that we've recently started, uh, export. And this can be combined with the Spike detection approach, something called spite band power. And the benefits of that approach are that you may be able to pick up something of from neon, said there, maybe too far way to be detected as a Spike, because the farther way you are from elector, the weaker that actual Spike way farm will look like I had of actual so you might be to pick up, you know, population of activity of things that are now maybe slightly outside the Normal recording radius.

What neuroscience is sometimes refer to as the hash of activity, the other stuff that's going on, uh, and you can look at through across many channels how that the background noises behaving and may be to get more juice out of the single that way. But IT comes at a cost. That signal is now a floating point representation, which means more expensive. Stand out of a power, means you have to find different ways to impress that, that are different than what you can apply to buy new instance. So with a lot of different chAllenges, social with these different models.

So also into communication, you're limited by the amount of data consent yeah and so and also because you are currently using the blue to protocol, if the bad stuff together, but you have to also do this keeping latency crazy low. I crazy anything to say about the latency?

yeah. This is a passion project to mind. So I want to build the best mouse in the world. I don't want to build like the the sheep is Sparky, whatever of elections. I want to build like the test, the roadster version of of a mouse.

And I really do think it's quite possible that within, you know five to ten years that most exports competitions are dominated by people with paralysis. This is like a very real possibility for number of reasons. One is that they're have access to the best technology to play video games effectively secondments have the time to do so. So those two factors together are particularly pot .

for uh export competitors unless uh people without clouds are also allowed to implant, which is IT is another way to interact the digital um device. And there's some there's something to that if if it's a fundamentally different experience, more efficient experience, even if it's not like some kind of fall on hide band with communication. If it's just the ability to move the mouse, uh, ten x faster, like the bits per second, if I can achieve a bits per second at ten x, what I can do with the mouse, it's a really interesting possibility what they can do, especially as you get really good at IT, uh, with training.

it's definitely the case that you have a higher ceiling performance like you because you don't have to buffer your intention through your ARM, through your muscle. You get just by nature of having a billion plant at all, like seventy five minutes I D A time on any action that you actually trying to take.

And there's some new answer to this, like there's evidence that the motor tex, you can out of plan out sequences of action so you might not get that whole benefit all the time. But first sort of like reaction times style, uh, games were used one up. Someone is over here, snipe on, you know, that kind of thing.

Uh, you actually do have just an inherent advantage because you don't need to go through the muscle. So the question is just how much faster can you make in and already you know faster than, uh, you know what you would do if you going through the mother from a latency point of view and learn the early stage of I think we can push IT. So of our underlines right now from brain Spike to crucial moments is about twenty two million seconds.

If you think about a the best mice in the world, the best game in mice, that's about five more second fish of agency, depending on how you measure, depending how fast screen freshes. A lot of characters tics matter there. But yeah, and the rough time for like a neuron in the brain to actually impact your a command of your hand is about seven, five, nine second.

So if you look at those numbers, you can see that we're already like you know competitive and slightly faster than what you would get by actually moving your moving your hand. And this is something that you know if you ask no. And about IT, when he moved the cursor for the first time, we asked him about this, something super curious about that.

What does that feel like when you're modulating, you know, a click intention, or when your trying to move the cursor to the right? He said that moves before he is like actually intending IT to, which is kind of a surreal thing and something that, uh, you know, I would love to experience myself one day. What does I like have a thing just be so immediate, so fluid that he feels like it's happening before you're actually tending to move?

Yeah, suppose we ve got used that lencs, that natural lencs that happens. Uh, so is the currently the bottle of communication. So like the blue est communication, is that what's the actual bottle? Like, I mean, I was going to be a bottle. The country.

yeah a couple things. So kind of hilariously bluetooth, uh, low energy protocol has uh some restrictions on how F I communicate. So the protocol self establishes is the andred of you know the most frequently of update. You can stand on the order of seven point five minutes seconds.

And um as we push the and see down to the level of sort of individual Spikes impacting control, that level of a resolution, that kind of project is going to become a limiting factor at some scale. Um another sort of important new on to this is it's not just not like itself that part of the equation. If you start pushing linseed so below the level of half a screen refresh, then you have another problem. You eat your whole system to be able to uh be as reactive as uh limits of what the technology can office like you need the scree like hundred twenty hours just doesn't know work anymore if you trying to have something respond something that know the level of one door. Second.

it's a really cool chAllenge. I also liked that for A T shirt. The the best mouse in the world.

Tell me on the receiving end. So the decoding step now we we figured out what the Spikes are, got them all together. Now was sending that over, uh, to the APP. What's the decoding step look like? Yeah so maybe first.

what is decoding? I think this part a lot of false listening that just no what .

what IT means to decode brain activity actually what is so there's there's an implant that's wirelessly communicating with any digital devices as an apple stalled. So maybe can you tell me how of what the ad is, what the softer is outside of the other brain?

yeah. So maybe working back from the goal. The goal is to help someone with policies, in this case, no land, be able to navigate his computer independently. And we think the best way to do that is to offer them the same tools that we have to navigate software r, because we don't want to have to rebuild entire soft ecosystem for the brain, at least not yet. Maybe someday, you can imagine the U X is are built natively for A B C I.

But in terms of what's useful for people today, I think we most people would prefer to be all that just control mouse and keyboard inputs to all the applications that they wanted use for their daily jobs, for a communicating with A A. And so the job of the application is really to translate this wireless stream of branded a coming off the implant into control of the computer. And we do that by essentially building a mapping from creativity to serve the H.

I. D. Inputs to the actual hardware. So H I is just the party call for communicating like input device events.

So for example, move mouse to this position or press this key done. And so that mapping is fundamentally what the office responsible for. But there's a lot of new ones of how that mapping works.

We spent a lot of time to try to get right, and we're still in the early stages of a long journey to figure out how to do that optima. So one part of that process, decoding. So decoding is the process of taking the statistical patterns of brain data that's being channel across the good of connection to the application and turning IT into, for example, a mouse women.

And the decoding step, you can think of IT in a couple of different parts. So some of any machine on any problem there is a training step, is an instance step. The training step in our case is uh, very uh, intricate behavioral, a process where the user has to imagine doing different actions.

So for example, they may be presented a screen with a curser on IT and y'll be asked to push that crusher to the right. Then imagine pushing the curse to the left, pushed up, pushed IT down. And we can basically build up a pattern, or using any such of modern m, uh, a mapping of, given this brain data and that imagine behavior map one to the other.

And then a test time, you take that same pattern matching system, in our case, is a deep neal network, and you run IT, and you take the life human brain day is coming off the implant, you decoded by patch matching to what you saw that calibration time. And you use that for a control. The computer now couple likes a rabid holes are I think are quite interesting.

One of them has to do with how you build that best temple matching system because there's A A variety of. Behavioral chAllenges and also debug in chAllenges when you're working with someone who's paralyse because again, fundamentally, you don't observe what they're trying to do. You can see them attempt to move their hand. And so you have to figure out a way to instruct the user to do something and validate that are doing IT correctly, such that then you can downstream build with confidence the mapping between the arrow Spikes and the intended action.

And by doing the action correctly, what I really mean is at the level of resolution of what neuron to doing so if in ideal world, you could get a signal of behavior intent, that is ground truth accurate at the scale of sort of one no second resolution, then with high confidence, I could build a mapping from my neuropathy, that behavioral attention. But the chAllenges, again, that you don't observe what they're actually doing. And so there's a lot of new ones to how you build user experiences that give you more than just sort of a course, on average, corrupt representation of what the users attending to do. If you want to build the world best mouse, you really want IT to be as responsible as possible. You wanted to be able to do exact the users attending at every sort of step along the way, not just on average, be correct when you trying to move left to right and building a behavioral or of calibration game, or our sort software experience that gives you that level resolution, is what we spent a lot.

So the calibration process, the interface, has to encourage precision, meaning, like whatever IT does. This should be super intuitive, that the next thing the human is going to likely do is exactly that intention that you need and only that intention yeah and you don't have any feedback except that may be speaking to you afterwards what they actually did you can ah so that's that's fundamentally that is really exciting. U X, because that's all on the U. X, is not just about being friendly or nice or usable yeah it's like user .

experience is how that works.

is how works yeah for the calibration and calibration, at least at this stage of newer link is like fundamental to the Operation of the thing in and not just calibration but continue calibration essentially yeah and maybe yeah .

you said something that I think is worth expLoring that a little bit. You said a primarily A U X chAllen and I think a large component IT is but there is also a very interesting machine learn in china er which is given some you know a dataset including some on average crack behavior of asking the user to move up or move down, move right left.

And given the data set of the old bikes is is there a way to win fur in some kind of si supervised or entirely on supervised way, what the high resolution version of their intention is? And if you think about IT, like there probably is because there are enough data points in the data set, enough constraints on your model that there should be away with the right sort of formulation to let the model figure out itself. For example, like this millisecond, this is exactly how hard they are pushing upwards. And at this most second, this is how hard they are .

trying to push upwards. It's really important. Have very clean labels, yes.

So like the problem, much harder from the mine. Learning perspectives of the labels are noisy. That's correct. And and then to get the clean labels that's A U X chAllenge correct.

Although clean label S I think maybe it's worth explain what that exactly means. I think any given labelling strategy will have some number of assumptions to make about what the users attempting to do. Those assumptions can be formulated in a loss function or they can be formulated in terms of heuristics that you might use to just try to estimate to estimate what the user are trying to do.

And what really matters is how accurate those assumptions. For example, you might say, hey, user, pushed upwards and follow the speed of this cursor and your heroic might be that they're trying to do exactly that Chris is trying to do, I know, competing hurry tics might be they're actually trying to go slightly faster to beginning in the movement and slides store at the end. And those competing theistical mayor may not be accurate or functions what they are trying to do.

Another version of the task might be, hay user, imagine moving this curse, a fixed offset. So rather than follow the cursor, just try to move IT exactly two hundred pixel to the right. So here's the cursor, here's the target.

okay? Cursor disappears. Try to move that now invisible cursor to other people to the right. And the assumption in that case would be that the user can actually modulate correctly that position offset, but that position offset assumption might be a weaker assumption, and therefore potentially you can make a more accurate.

Then these physics are trying to estimate each millisecond what the user are trying to do. So you you can imagine different tasks make different assumptions about the the nature, the use of intention. And those assumptions being correct is what I want .

to think of is a clean label that that would be visualizing this a cursor, and you want to move the cursor the right to left up and down, or maybe move them by a certain offset. So that's one way that the best way to calibration. So for example, an alternative crazy way that probably playing a role. Here's a game like a great where you're just getting a very large amount of data. The person playing a game where if they're in the state of flow, maybe you can get cleaned signal as a side effect yeah is that or is that is that not effective way for initial collaboration?

Yeah great question. There's a lot to impact there. So uh, the first thing I would dry distinction between this sort of open loop, first close loop. So open, look, what I mean by that is the user is so of going from zero to one.

They have no model at all and they're trying to get to the place where they have some level of control at all in that set up. You really need to have some task that gives the user a hint t of what you wanted to do, such that you can build this mapping again from brain data to to open. Then once they have a model, you could imagine them using that model and actually adapted to in and figuring out the right way to use IT himself and then retraining on that data to give you A A boost and performance.

There's a lot of chAllenges to say with both of these techniques, and we can of rabid hold the both and if if you're interested. But the chAllenge with the open move task is had the user themselves doesn't get proper a suspect feedback about what they are doing. They don't you know necessarily perceive themselves or fuel, you know the mouse under their hand when the using and open when they are going to do and open the calibration, they are being asked to perform something like, imagine if users are whole right ARM numbed and you stuck in a box and you couldn't see IT.

You're no visual feedback and you had no propose of the feedback about what the position or activity of your aren't was. And now you're asked, okay, given the thing on the screen that's moving from up to right, match that speed and you basically can try your best to, you know, invoke whatever that imagined action is in your brain that's moving the cursor from left, right. But in any situation you gonna be inaccurate and maybe inconsistent, and how you do that task.

And so that sort of the fundamental chAllenge of open loop, the chAllenge with close loop is that once the users given a model and they're able to start moving the mouse on their own, they're going to very naturally adapt to that model. And that could adaptation between the model learning what they're doing and the user are learning how to use the model may not find you the best of global minimum. And maybe that your first model was noisy in some ways or um you know maybe just had some like york, there's some like part of the data strip tion.

I didn't cover super well and the user now figures out because there are you know bRiley user like no one, they figure out the right sequence, imagine motion, motion or the right angle left to hold their hand out to get IT to work. And they'll get IT to work, great. But then the next day, they come back to their device and maybe they don't run exactly all the tricks that they use in the previous day. And so those are complicated. So the feedback I call here that can now that can emerge and can make you a very, very difficult debug in process.

okay. There's a lot of really fascinating things there. You actually just to stay on the on the close loop I have as in situations this actually happened watching, uh, psychology, they use piece of software and they don't know how to program themselves.

They use peace soft somebody and has a bunch of bugs. And they figure out like, and they ve been using IT for years, they figure out ways to walk around. That just happens like nobody nobody like consider maybe we should fix, they just adapt.

And that's a really interesting notion that we just really, really good at adapting. But you need to still that might not be the overall yeah okay. So how do you solve that problem? You have to restart from scratch every once a well kind of thing.

Yeah is a good question. Um first, informal. I said this is not a solve problem. And for anyone who's you know listening in academia who works on B, C, I would also say this is not a problem that solved by simply scaling channel count.

So this is, you know, maybe that can help, and you can get the richer covering structures that you can use to exploit when trying to come up with a bling charge. Es, but if you know you're interested in problems that aren't going to be solved inherently by scaling channel t, this is one of them. yeah.

So how do you solve IT? A, it's not a sole problem that the first thing I want to make sure gets across. The second thing is any solution that involves close, close loop is going to become a very difficile debug in problem.

And one of my sort of general horis tics for choosing what prompt attack is that you wanted, choose the one that can be the easier to debug. Because if you can do that, uh, even if the ceiling is lower, you gonna be able to move faster because you have a tighter iteration loop debug in the problem. And in the open loop setting, there is not a feedback cycle bug with the user in the loop.

And so there is some reason to think that, that should be an easier debuting problem. The other thing that work understanding is that even in the close upsetting, there is no special software magic of how to infer what the user is truly attempting to do in the close of setting, although they're moving the curse on the screen. There maybe attempting something different than much model is outputs.

So what the model is outputs is not a signal that you can use to retrain if you want to able to improve the model further. You still have this very complicated estimation or unsupervised problem of figuring out what is the true user intention on underlying that. And so the opening problem has the nice property of being easy to debug.

And the second nice property of IT has all the same information content as a close of scenario. Everything I want to I want to mention and call out is that this problem doesn't need to be solved in order to give useful control of people. Um you know even today with the solutions we have now and the academy has built up over over decades the love of control that can be given to a user you know today is quite useful.

IT doesn't need to be solved to get to that develop control. But again, I want to build the world's best mouse. I want to make IT you know so good that is not a question that you want IT. And to build the world's best mouse, the superhuman version, you really need to a nail that problem in a couple maybe details of previous studies that we've on internally that I think are very interesting to understand, thinking about how to solve this problem.

The first is that even when you have ground true data, what the users trying to do, and you can get this with an able body monkey, a monkey that has an early device and planted and moving a mouse to control a computer, even with that ground true data, 3IT turns out that the optimal thing to predict to produce high performance B。 C. I is not just the direct control of the mouse.

You can imagine, you know, building the data set of what's going on in the brain, and what is the mouse exactly doing on the table. And IT turns out that if you build the mapping from neurobiology dict, exactly the mouse is doing that model will perform worse then a model that is trained to predict some of high love assumptions about about the user might be trying to do. For example, assuming that the monkey is trying to go industry line to the target, IT turns out that making those assumptions is actually a more effective in producing a model than actually predicting the .

underline handles the intention, not like the physical movement, whatever. There's a obvious a very strong correlation between the two and a more powerful thing to be chasing, right? Well, that that's also super interesting.

I mean, the intention itself is fascinating because yes, with the B, C, I here, in this case with the digital telepathy, you're acting on the intention, not the action, which is why there's an experience of like feeling like this happening before you meant for that to happen, that is so cool. And that is why you could achieve like super human performance problem in terms of the control amount. So the four open loop, just to clarify. So whenever the a person is test, like move the mass to the right, he said there's not feedback .

so they didn't .

get to get that satisfaction of like actually getting IT to move.

right？ So you you could imagine giving the user feedback on a screen, but uh, it's difficulties at this point. You don't know what they're tempting to do.

So what what can you show them that we basically give them a single love on doing this correctly or not correctly? So IT take is a very specific example. If maybe your calibration task looks like you're trying to move to curse her a certain position offset so your instructions to the user. R hey, the curses here.

Now, when the cursor disappears, a manager moving at two hundred pixel from where I was to the right to be over this target in that kinda era, you could imagine coming up with some sort of consistency metric that you could display the user of, okay, I know what the spite train looks like, an average. When you do this action to the right, maybe I can produce some of the probabilities of estate of how likely is that a to be the action you talk? And given the latest trial or rejected that you imagined, and I can give the user some sort of feedback of how consistent are they across different trials. You could also imagine that if the user is prompted with that kind of uh, consistency metric, that maybe they just become more behaviorally engage to begin with because the task is kind of boring when you don't have any feedback at all. And so there may be benefits to the you know the user experience of showing something on the screen, even if it's not accurate, just because we give the user motivated to try to increase that number pushing upwards.

So there's a psychology element here.

Yeah, absolutely.

And again, all of that is U. X. chAllenge. How much signal drift is there over to hour, day to day, week to week, months a month? How often you have to recover bate because of the signal drift?

Yeah uh so this is um a prom ve worked on both with N H P, not human primates before our clinical trial and then also with no land during the finanical trial. Maybe the first thing the world stating is what the goal is here. So the goal is really unable the user to have a plug and play experience where, I guess, like anything in but a play experience where they have you can use the device whenever they want to, however they want to.

And uh, that's really what were aiming for. And so there is can be a set of solutions that get to that state without um considering the stationaries prom. So maybe the first solution here that's important is that they can recover ber whatever they want.

This is something that that no one has you able to do today, so you can recognize the system, you know, two A M in the midst night without his know caretaker or parents or friends around to help push a button for him. The other important part solution is that when you have a good model calibrated, that you can continue using that without any interview calibrate. So how often he has to do this recognition take depends really on his appetite por performance.

There are a we observe a serve a degredation through time of how well any individual model works. But this can be mitigated behaviorally by the user adapting their control strategy. IT can also be medicated through a combination of the software features that we we provide the user.

For example, we let the user adjust exactly how fast the cursor is moving ah we call that the game, for example, the game of how fast the cursor reacts to any given input intention. They can also adjusted smoothing how smooth the output of that cursor intention actually is. That can also just the friction, which is how easier to that to stop and holds dell and all these software tools allow the user a great good deal of flexibility and trouble shooting mechanism will .

solve all this is done by looking to the right side of the screens, selecting the mixer. And the mixer you have is the D J .

mode DJ mtc. I mean.

it's a really well done interface. It's really, really well done. And so yeah there's that bias uh that there's a cursor drift that no one talked about in a stream.

Uh, although he said that a you guys were just playing wrong with IT with him and constantly improving. So that could have been just a snapshot of that particular moment of particular day. But he said that there was a this cursor drift in his bias that could be removed by him. I guess, looking to the right side screen and less side of the screen to kind of adjust bias. Yeah that's one interface action, I guess.

So the there there is some prior work with uh to bring a clinical trial participants where they ponder this idea. Bias correction the way we've done, I think is very prototyped very uh uh beautiful user experience where the user can essentially a flash the curse over to the side, the screen and IT opens up a window where they can actually uh of adjust or tune exactly the the biased of the curse. So bias made for people aren't familiar is just sort of what is the default motion is the cursor, if you imagining nothing and IT turns out that that's one of the first um first of qualia of the critical control experiences .

is impacted by neurosis .

I mean I don't know how I to describe IT like you not .

the not the way of the world the first experience yeah I mean it's IT sounds polite, but IT is a deeply true. There is an experience when IT works well. IT is a joyful, a really pleasant experience.

And when IT doesn't work was a very fresh ate experience. That's like the r of U. X. It's like you have the possibility to frustrate people or the possibility to give them joy.

And at the end of day, really is true. The case that U X is how the thing works. And so it's not just like what's join on the screen is also know what control surface is that a decoder provide the user we want them to feel like there in the f one car.

Not like the you some like many fan right you and that's really truly is how we think about IT. Um no one himself as an next one fan. So um we refer to ourselves as a big crew. He really is truly the f one driver and there is different no control surfaces that that different kinds of cars and airplanes provide the user.

Um we take a lot inspiration from how we designing, how the cursor should behave and what may be when you once of this is you know even details like when you move a mouse on a macbook track pad, the sort of response curve of how that uh input that you give the track pad translates to curse movement is different than how IT works with a mouse. When you move on the track pad, there's a different response function, different curve to how much a movement translates to input the computer, then when you do physically with a lot. And that's because somebody SAT down a long time ago when their design initial inputs sums to any computer and they thought through exactly how uh, IT feels to use these different systems.

And now we're designers of the next generation of this and put this to a computer which is entirely done by the brain, and there is no purpose cept of feedback. Again, you don't feel the mouse in your hand, you don't feel the keys under your fingertips, and you want a control surface that still makes IT easy, intuitive for the user who to understand the state, the system and how to achieve what they want to achieve. And ultimately, the end goal is that that U.

X is completely pads in the back and have become something that's so natural, intuitive and so conscious to the user. And they just should feel like they have basically direct control over the the cursor or just does what they wanted to do. They are not thinking about the implement mentation of how to make you do what they wanted to do is just doing what they wanted to do.

Is there some kind of things along the lines of like fits law, where you should move the mouse in a certain kind of way that maximizes your chance to hit the target and even know what i'm asking, but i'm hoping the intention of my question will land on on a profound answer. No h, is there some kind of understanding of the laws of U X. When IT comes uh to the context of somebody using their brain to control IT like that's different than actual with a mouse.

I think we're in the early stages of discovering those laws, so I wouldn't claim to have solved that problem yet. Uh, there's definitely some things we've learned that uh make IT uh easier for the user to get stuff done and it's pretty straight d when you verbalized IT but takes a while to actually get to that point when you in the process of debug ing the stuff in the trenches. One of those things is that the any any machine learning system you build has some number of areas. And uh, IT matters how those errors translate to the downs in music experience.

For example, if you developing a search algorithm in your photos, if you search for, you know your friend joe and IT, pull up a hot of your friend Joseph e maybe that's not a big deal because the cost of an error is not that high in, uh, a different scenario where you're trying to you know detect insurance father something like this and you directly sending someone to court because of some machine army model up. But then the areas make a lot more. Seems to be careful about you might be very thoughts about how those areas translate to downing am effects.

The same is in B, C, I. So for example, if you're building a model that decoding of velocity output from the brain versus an output where you're trying to module the left click, for example, these have to have a different tradeoffs of how precisely to be before IT becomes useful to the and user. For pho city is okay to be on average crack because the output, the models integrated to time.

So if the users trying to click a position a and the current their position b, they're trying to navigate over time to keep between those two points. And as long as the output of the model is on average craft, they can sort of steer through time with the user control loop in the, in the mix they can get to the point they want to get to. The same is not of a click.

For a click, you performing IT almost instantly and the scale of you know their inspiring. And so you want to be very sure that that click is correct because of false click can be very destructive. The user you might also close to have, they're trying to you do something in to lose all their progress.

They might accidently like you know hit some sand button and some text that is only like half composed and read funny after. Ah so you there's different sort of cost functions associated with theirs in this space. And part of the U. S. Sign is understanding how to build a solution that is when it's wrong.

still using the end user so fast assigning cost to every action when uh, an era occurs. So every action, if an error has a certain cost and incorporating that into how you interpret the intention, mapping IT to the action is really important. I didn't quite until you said IT realized there's a cost to like extending the text early. It's like a very expensive cause .

it's per on if you actually like if you're cursor r imagine if you're cursor miscast p every once in a that's like superman noxious. And the worst private is usually when the users trying to click, they're also holding still because they're over the time they want hit and they're ready to, which means that in the data sets that we build on average is the case that serve low speed or desire, hold still coral with when the users are jumping to click.

Well, that is really fascinating.

It's not the case. You know, people think that click is a binary of this must be super easy to decode. Well, yes, IT is. But the bar is so much higher for to become a useful thing for the user.

And these ways to solve this, I mean, you consider to take the compound approach of, well, let's just give the like, let's take five seconds to, let's take a huge window of times, we can be very confident about the answer. But again, world's best mouse, the worlds best mouse, doesn't take a second to click or five and second like that takes five minutes, seconds to take or less. And so if you're aiming for that kind bar, you really want to solve the line.

So maybe is a good place to ask about how to measure performs this whole bits per second. What um can you like explain what you mean by that may be a good place to start us to talk about web grades a game as a good illustration of the measurement of performance.

Yeah maybe i'll take one zoom out step there, which is just explaining why ah we care to mature this at all. So again, article is to provide the user the ability to control the computer as well as I can and hopefully Better. And that means that they can do IT at the same speed as what I can do.

IT means that they have access to all the same functionality that I have, including, you know, all those details, I command tab, command space, you know, all of this stuff. And you will do IT with the brain. And with the same level of reliability is what I can do with my muscles.

And that's a high bar. Until we intend to measure and quantify if we asked to that to understand how progressive towards that goal, there's many ways to measure be P S by this isn't the only way. But um we present the user a grave targets and basically we compute uh a score which is dependent on how fast and actually they can select and then how smaller the targets.

And the more targets are on the screen, the smaller that they are, the more information you present per click. And so if you think about IT from information theory point of view, you can communicate across different information to elect channels. And one such channel is a typing interface.

You could imagine that built out of a grid just like a soft y board on the screen. And uh, bits per second is a measure computed by taking the log of the number of targets on the screen. Uh, you can sub try to one if you care to model keyboard because you have to subtract one for the delete on the keyboard.

But log of the number target on the screen times the number of correct elections, minus incorrect, divided by some time went on, for example, sixty seconds. And that through the stand way to measure a access, the control task, dema and all credit. And the world goes to this professor di of stanford, who came up with that task.

And he's also one of my inspirations for being in the field. So all of the country in the world to him for coming with a standardized metric to facilitate this kind of bragging rights that we have now, how to say that? No one, this is the best in the world.

That is task with B. C. I. It's very important for progress that you have ized matters that people can compare across different techniques and approaches.

How well does this do? So do s him to all all the team of stanford. Um yeah so for newland and for me, plane this task, there's also different modes that you can figure this task.

So the web good task can be presented as just sort of a left luck on the screen. Or you could have, you know, targets that you just do all over, or you have targets that you laugh right clack on. You can have targets that are left right click, middle ick, scaling, clicking and drug.

You do all sorts of things within this, this general ramework but as a simplest curious farm is just blue targets drop on screen. Blue means left luck. That's a simplest farm of the game.

And uh the prior record here in uh academic work and uh narrow link internally with service hp uh have all been matched or beaten by by no one with his narrow ling device. So a prior true nearly the sort of world record for human using device is a somewhere between four point two to four point six. Pps and exactly what paper you read, how interpreted a knowledge current record is a point five B P S, and again, the sort of media nearly inc.

Performance ten B P S. You can think of IT roughly, as is eighty five percent to level of control of medium neural linker using their current to slide blue targets on the screen. And um yeah I think there is a very interesting journey head to get us to that same level of ten B P S.

Performance is not the case. Start of the trick, psychotropic, you know four to six P P S and then six to A P P S are going to be the one to get us from eight to ten. And in my view, the court chAllenger is really the labelling problem. It's how do you understand at a very, very fine resolution what the users attempting to do? And yeah, I highly encourage focus academia to to work on this problem.

What's the journey with knowing on that quest of increasing the B. P. S? On web grade in march, he said that he selected eighty nine thousand and two hundred and eighty five targets and web grade.

Yes, so he loves this game. He really serious about improving his performance in this game. So what is the journey of trying to figure out how to improve that performance? How much can I be done on the decoding side?

How much can I be done on the calibration side? How much can I be done on the known side of like figuring out how to convey his intention more cleanly? Yeah, no, this is a great question.

So in my view, one of the primary reasons why no one is performance is so good is because of no land. No one is extremely focused and very energetic. He will play web grade sometimes for like four hours in the middle night, like from two hundred and six who we playing website just because he wants to push you to the limit when we can do.

And you know, this is not us, like ask him to do that. I want to be clear that we're not saying you should play web gate tonight. We just gave him the game as part of a research, and he is able to play independently. Practice whatever he wants and he really pushes hard to push IT.

The technology is the episode limit and he views us like you know his job really to make us be the bottle and boy has he done that well um and so that the first knowledge is that you know here's extremely motivated to make this work. Have also had the pledge to me other you know chemical cp participants from range in other trials and they very much share the same attitude of like they they view this as their life's work to uh you know advanced technologies as much as I can. And if that means like targets on the screen for four hours two m to six am, then so will be in uh there's something extremely animal about that that's worth uh calling out OK.

So now how do you how do you sort to get from where he started, which is no crucial control ID. P P. S.

So I mean, when he started, there's a huge amount of learning to do on his side and our side to figure out, uh, what's the most actual control for him and the most intuit control for him is, uh, if you have to find the set intersection of what do we have the signal to decode so we don't pick up, you know, every single learn in the motor k protect, which means we don't have a presentation for every part of the body. So there maybe something that we have Better decode performance on the others. For example, on his left hand, we have a lot of difficulty.

This has left ring finger from his left the finger, but his right hand, we have a good, no good control and good modulation detected from the neurons able to record for his pinking stump and is an next figure. So you can imagine how these different um you know spaces of moderate activity into with what's the most intuit for him. And this is even all of our time. So once we gave him the ability to calibrate models on his own, able to go to explore various different ways to imagine and control in the cursor, for example, he can imagine control in the cursor by within his rests on side, or by moving his entire ARM by one pointed to his feet. You know, he tried the whole bunch of stop to explore the space of what is the most natural way for him to control the cursor that at the same time is easy .

for us to be clarify. Its through the body mapping procedure, you're able to figure out which finger he can move uh yes .

yes that's one way to do IT um maybe one new ones of uh, when he's doing him, he can imagine many more things and we represent that visual on this thing so we show him serve out strictly here's a cursor. You figure out what works the best for you and we obviously hints about what will work best from the body. My procedure of, you know, we know that particular action we can represent well, but truly up to him to go and expLoring .

and figure out what works the best. But at which point does he no longer visualized the movement of his body and socialize the movement of the? How quickly does he go from? Quickly does he get there?

So this happened on our tuesday. I remember this day very clearly, because at some point during the during the day, uh, IT looked like he wasn't doing super well, like look like the model wasn't performing super well and he was like being distracted. But he actually wasn't the case.

Like what actually happened was he was trying something new where he was just control the cursor so he wasn't imagining moving his hand anymore. He was just imagine, I don't know what I would like abstract intention to move the cursor on the screen and ah I cannot tell you what the difference to in those two things are. I truly cannot he is just to me before I cannot um you know give a first person account what that's like. But the exclusives that he ordered in that moment where you know enough to suggest that is a very quality, timely, different experience for him to just have direct neural control over a curse.

I wonder if there's a way through U. X to encourage a human being to discover that because he discovered IT, like he said, uh, to me, that he's a pioneer. So he discovered that in his own, through all of this, a the process of trying to trying to move the course, there are different kinds of intentions. But that is clearly a really powerful thing to arrive at, which is to let go of trying to control the fingers and hand, and control the actual digital device of your mind.

that you access how IT works and are the ideal you access one, that is, the user doesn't have to think about what they need to do in order to get IT done. They just, I just does IT as so .

fascinating. But I wonder on the, on the biological side, how long IT takes for the brain to adapt? Yeah so is is just simply learning like high level software or is there like in your plasticity component where like that the brains adJusting slowly?

Yeah the truth I don't know. Um i'm very excited see with through the second participant that we implement what you know what the journey is like for them because we'll have learned a lot more potentially we can help them understand, explore that distraction more quickly. This is something I didn't know.

This wasn't me perfect, prompting no one to go try this. He was just expLoring how to use his device and figure IT out himself. But now that we know that that's a possibility that maybe there's way to, for example, hint the user don't try super hard during collaboration to do something that feels natural or just directly control the curse, you know, don't iggy? Business action.

And from there, we should be uh, hopefully understand how this is for somebody who has not experience that before. Maybe that's the default mode of Operation for them. We have to go to this immediate phase of explicit .

motions or maybe if that naturally happens for people, you can just occasionally encourage them to allow themselves to move the curse, right? Actually, sometimes, just like the form and a mile, just acknowledge .

that that's possible. Push.

you do IT yeah, enables you to do IT and then becomes trivial. And that also makes you wonder, is the cool thing about humans. If once there is a lot more human participants, they will discover things that are possible.

yes.

and have a and share other and that because of them sharing IT, they'll be able to do IT. Um the also that's that's unlock for everybody yeah because just acknowledge sometimes is the thing that enable them to do IT.

Yeah just coming on that too. Like there's we've probably tried like a thousand different ways to do various um aspects of decoding and now we know like what the right subspecies to continue explain for. Again, thanks to know the many hours is put into this.

And so even just that help, like help constraints of the beam search of different approaches that we could explore, really help IT for the next percent. You, the set of things will get to try and day won, how fast we hope to get them the useful control, how fast we can able to use IT independently and to give value the system. So yeah, massive hats off to to no london and all the participants is that came before form, uh, to make this technology really.

So how often of the of the coder? no. And mention like, okay, there's a new update that we're working on and that in the stream, he said he plays the snake game because it's like super hard.

It's a good way for him to test like how good update ata. So and he says like sometimes the update is A A A backwards as it's a constant like iteration. So how often like what is the update in tail is the most other side? Yeah couple. So what is quite worth .

trying distinction between research sessions where we're actively trying different things, understand like what the best approaches is very sort of independent youth were want to have, yes, no ability just go use the device how anybody would want to use their macbook.

And uh, so he's referring is, I think usually in the contest of a research session where we're trying you know many, many different approaches to you know even until to come up with and more accurate coding. And in those seniors, I mean, we try in any given session, he sometimes work for like eight hours a day. And so that can be no hundred ds of different models that we will try in that day like a lot of different things.

Um now it's also worth noting that we update the application he uses quite frequently. I think sometimes up to like four, five times a day will update his application with different features or bug fixes or feedback that he's given us. So he's been able to he's a very ticula person who is a part of the solution.

She's not a complaining person, he says he here is the thing that i've i've discovered is not often what my flock here. Something is how to fix. Let me know what you thought.

Sorry, let's figure out how to often and IT often happens that those things are addressed within no, a couple hours of him giving us this feedback. That's a kind of federation cycle have. And so sometimes began in the session to give us feedback at the end session. He he's giving us feedback on the next election of that of of that processor .

that set up because one of things you mention that there is two hundred seven one pages of notes taken from the B C. I. sessions. And this was just in march. So one of the amazing things about human beings that they can provide, especially once who are smart and excited, and all like positive and good voice, like no, and they can provide feedback, continuous feed back.

yeah. IT also requires a just a brag on a team a little bit. I work with a lot of exceptional people, and IT requires the team being absolutely laser focused on the user and what will be the best for them.

And IT requires like a level of women of, okay, this is what the uses feedback g was. I've all these meetings we're going to skip to that today. We're going to do this. You know that level of focus commitment is A I would say, and unappreciated in the world and also uh you know you actually have to have the talented will execute on these things effectively. And um yeah we have that .

in in loads yeah and this is such a interesting space of U. S. Design because you have there's so many unknowns here.

And I can tell U. X is difficult because of how many people do IT poorly. It's just not a trivial thing.

Yeah it's also, you know U X is not something that you can always solved by just constant international on different things like sometimes really to step back and think globally. And I even like the right of minimal to be tracing down for a solution. Like there's a lot of problems in which sort of fast iteration cycle is uh, the predictor of how successful you will be.

As a good example, like in an R, L simulation, for example, more frequently you get reward, the faster you can progress is just an easier learning problem, the more frequently you get things back. But U, X is not that way. You mean users are actually quite often wrong about what the right solution is. And IT requires a deep understanding of the technical system. And what's possible, combine with what the problem is, are you trying to solve not just how the user express to you for what the true underlying problem is actually go to the right place?

Yeah, that's the old like stores of Steve jobs, like growling in there, like yet. The users of goods is a useful signal, but it's not a perfect signal, and sometimes you have to remove the flap. He is travel over the, I got all the crazy stories of their jobs, like making wild design decisions.

But there are some some of this aesthetic that some of IT is about the love you put into the design, which is very much a Steve jobs johna type thing. But when when you have a human being using their brain to interact with ate, there is also is deeply about function. It's not just aesthetic and that you have to empathy with with the human being before you.

While not always listening to them directly, they get to deeply empathize. It's fascinating. It's really, really fascinating.

And at the same time, iterate right, but not iterate in small ways. Sometimes I complete like we building the design. He said that no one said the early days the U.

S. Sucked, but you improve quickly. IT was that journey like.

yeah, I mean, i'll give one concrete example. So he really want to build the read maner something that he I mean, yeah, that sounds like a simple thing, but actually a really big deal for him and he couldn't do IT with this moustier just IT wasn't accessible that you can't score with mouse c on his ipad and on the website that you want two pills used .

to read the the nest manager and so maybe a good quick post to say the math stick is a thing is using holding a stick in his mouth to screw on a tablet right?

Yeah basically you is the styles that you hold the .

to and IT hurts and it's inefficient yeah and maybe it's also worth .

coming out. There are other alternatives of technologies, but the particular situation, knowledge and this is not uncommon and I think is also not well stood by folks, is that you know he's relatively we'll have muscles fast from time to time.

And so any assistance of technology requires into a position drink from from camera for an example nightmare er or anything that requires to put something this mouth, this is a no go because he'll either be shifted out of frame when he has spasm or if you have something mouth is seven in the face in these thousand too hard. So these kind of considerations are important when thinking about what advantages of P. C.

I has in someone's life, if IT, if IT fits organometallic life in a way that you can use IT independently when your caretaker is not there, wherever you want to, either in the bed or in the chair, depending on, you know, your comfort level and your desire to have pressure source. You know, all these factors matter a lot in how good the solution is in that users, in that users life. So one of these very fun examples score.

So again, many guys, something he wants to be able to read. And a, there's many ways to do score with A, B, C, I, you can imagine, like different gestures, for example, the user could do that would move the, move the page. But the girls are very fascinating.

A control surface, because it's a huge thing on the on the on screen in front of you. So any sort of gitter in the model output, any sort of air in the model output causes like, uh, an earthquake on the screen like you really don't want to have your minger page. Are you trying to read be shifted up and down a few pixel just because you know your score decoder is not completely accurate.

And so this was an example where, uh, we had to figure out how to format the problem away, that the errors of the system whenever they do a car and we will do our best to minimize stem. But whenever those area do occur, that IT doesn't interpret the quality. Again, the experience that the users is having that does not interpret the flow of reading their book.

And so what we ended up building is this really brilliant um feature. This is intimate jews who worked on this really brilliant work called quick grow and quick roll basically looks at the screen and IT identifies where on the screen or score bars and IT does this by deeply integrating with macos. Understand where are the the score bars actively present on the screen using the sort of accessibility to that available to a maps.

And we identified where that those copies are and provided A, B, C, I score bar. And the B, C, I score bar looks similar to a Normal scrawl bar, but IT behaves very differently in that one user of move over to a you curse sort of worth onto IT sort of attached or loches onto IT. And then once you push up or down in the same way that you use a push to control you, the Normal cursor IT actually moves the screen for you.

So like remap in the velocity to a scroll action. And the reason that feels so actually intuitive is that when you move over to attach to IT, a feels like magnetics you like a stuck into IT. And then is one continuous action you don't want to, like, switch your imagine movement you sort of snapped on to, and then you could to go, you just immediately can start pulling the page down and push IT up.

And if he wants you get that right, there is so many little nuances of how this co behavior works to make IT natural tuition. So one example is momentum. A when you score a page as your fingers on screen, you know you you actually have some like flow, like IT doesn't stop, right? When you lift your finger up, the same is true with B, C, cools.

So we had to spend some time to figure out what are the right new answer. You don't feel the screen under your fingertip anymore. What is the rights of dynamic or what's the right amount of page give, if you will, uh, when you push you to make IT flow the right amount for the user to A A natural experience reading a book. And there's a million I mean, this I could tell you, you like there's so many little munshi, how exactly that will work that we spent probably like uh, a month getting right to make that feel extremely natural and .

uh uneasy for the user to and even the score and a smart phone with the finger feels extremely natural and pleasant and IT probably takes a extremely long time to get that right.

And actually the same kind of visionary uh U X design that we're talking about, don't always listen to the users but also listen to them and also have the visionary big like throw everything I would think from first principles but also not yeah yeah by the way, this makes me think that score bars on the desktop probably have stagnated and never taken that like because the snapped as snapped a good snap to school bar actually you're talking about is something that could potentially be extremely useful in the desktop setting. You even just for users to just improve the experience because the current goal by experience in the next top is horrible yeah is hard to find, hard to control. There's not a moment there's um and the intention should be clear when I start moving towards the score bar, there should be a snap into score bar action.

But of course, you know maybe A O pain that cost, but this hundreds of millions of people pain that cost on stop but anyway uh but in this case this is necessary because that an extra cost paid by knowing for the jitteriness. So you have to switch between the squalling and the reading. There has to be a face shift between the two, like when you're scaling, you're scaling, right, right?

So there is one drawback of the current the court approach, maybe each other just sort of case study here. So again, you ask is how IT works. And we think about that holistically from like the even the future detection level, what we detect in the brain, to how we design the decoder, what we choose to decode to then how IT works once being used by the user.

So not a good example in that of how IT works wants to actually using the decoder. Uh, now the output that display on the screen is not just what the decoder says. It's also a function of, uh, you know what's going on on the screen.

So we can understand, for example, that you know when you're trying to close a tab that very small, stupid little acx is extremely tiny, which is hard to get precisely hit. If you're dealing with serve a noisy output of the decoder, we can understand that that is a small al lex you might be trying to hit. And now you you make IT a bigger target for you, similar time when you typing on your phone.

If your are, you know, use like the I O S keyboard, for example, he actually adapt the target size of individual keys based on underlying language model. So it'll actually understand that i'm typing, hey, i'm going to see l IT will make the equi bigger because that those lux as a person going to go see. And so that kind of know predictin ess can make the experience much more smooth even without improvements.

The underline decoder or uh or future detection part attack. So we do that with a future called my etic targets to we actually index the screen. We understand OK. These are the places that are you know very small targets that might be difficult to hit. Here's the kind of current dynamics around that location that might be indicated of the user is let's make IT easier. Let's blow up the size of IT in a way that makes IT easier for the user to sort snapped on to that target to all these little details, they matter a lot in helping the .

user be independent in their day day living. So how much of the work on a decoder is generalized to p two, p three before p five, P M? How how do you improve the court and away as generalizable?

Yeah, great question. So the underlying signal were trying to decode is going to be a very different to than A P one. For example, channel number three hundred and forty five is going to mean something different in user one than the one user to, just because that elector that corresponds to channel forty five is gonna in next to a different neuron in user one person user to.

But the approach is the methods, the user experience of how do you get the right through the behavioral pattern from the user to associate eros inal. We hope that will translate over, over multiple generation of users and beyond that is very, very possible. In fact, quite likely that we've over fit to of newlands user experience, desires and preferences. And so what I hope to see his hat, uh you know when we get uh second, third forth participant that we find sort of what the right wide minimum are, then cover all the cases that make IT more interior everyone and hope fully there's across foundation of things where, oh, we didn't think about our this user because, you know, they can speak. But with this user who who just can fundamentally not speak at all, this user experience is not optimal and that will actually those improvements will we make there, hopefully transact that even people who can't speak, we don't feel comfortable doing so because .

we're a public setting like the doctor's office. So the actual mechanism of open loop labelling and then closely blawing will be the same and hopefully can generalize across the different users as they do the calibration step and the calibration that was prety cool. I mean, that in itself that the interesting thing about web grade, which is like close, it's like fun. I love IT when there is like this should be kind of idea of human competition, which is using actions of human would want to do anyway to get a lot of signal from you and I web. Is that like a nice video again, that also serves as Green calibration?

So funny. This is, i've heard this reaction so many times before, sort of, uh, you know, first user was implant. We had an internal perception that the first you will not find this fun. And so we thought really quite a bit actually about actually build other games that like are you know more interesting for the user so we can get this kind of data and help facilitate research that for long duration of this turns out that like people love this came I always loved IT, but I didn't know that there was a shared .

perception yeah just in case is not clear what good is. There's A A great of the, say, thirty five by thirty five cells. And one of the lights are blue. You have to move your mouse over that and click on IT. And if you miss IT.

it's red. And I for so many hours, so many hours.

And what's your record?

He said my, I think I have the highest at and now my record seventeen B P, S seven about if you imagine that thirty thirty, you about a hundred trials per minute. So one hundred crack selection in that one minute window, see your averaging about, know between five hundred, six and eight seconds per section.

So one, one of the reasons that I think I struggle that game is such a keyboard person. Everything is done with the a keyboard as if I can avoid touching the mouse is great. So how can you explain your high performance?

I have like a whole ritual I go through. So this is actually like a diet plan as so it's a whole thing. So the first thing has to fast.

For five days, I have to go to the mouth.

actually. I mean, the faster thing is important. So this is .

like you so what .

I do is I I actually I don't eat for little bit and and then i'll actually like a tonopah about her before I put and I this is a real thing. This a real thing yeah and then IT has to be really late night. This is again, the night I think we share.

But IT has to be like, you know, midnight two. I am kind of time window. And I have a very specific like physical position sidin, which is, uh, I used to be our homeschooled growing up and so I did most my work like on the floor, just like in my bedroom, whatever. And so I A very specific situation on the floor, on the floor, I said and play and then you have to make sure like there's not a lot of weight on your elbow replying so you can move quickly and then turn the gain of .

the curse to see the cursor weight up.

It's like small motions that so my rist is almost completely still.

I'm just moving my fingers yeah you know those just in a small attack anger the which i've been meaning to go on this rabid hole of people that um set the world record and tetris, those folks they're playing. There's a there's .

a way to do did you see .

this all move? Yeah you can find a way to do IT where like it's using a loophole like a bug that you can do some incredibly fast stuff. So it's along that line but not quite. But you do realize they'll be like a few programmer. I now listen to this cool.

fast and eat A T yes, please try my record. I, me, the reason I did this literally was just because I wanted the bar. Be I the number that we aim for us should not be like the media informing you, should be like this should build to beat all of us at least like that should be the minimum bar.

What do you think is possible like twenty?

Yeah, I don't know if the limits I mean the limits you can calculate just in terms of like screen new fresh rate and like current immo jumping to the next target. But there's I mean, i'm sure there's limits before that with just a reaction time, individual perception and things like this. Um i'd guess it's in the below forty, but above twenty somewhere in there probably there right there are never thinking about IT also matters. I got difficult task is you can imagine like some people might be able to do like ten thousand targets on the screen and maybe they can do Better with that way um so they sound like task accumulation ation you can do to try to boost your performance as well.

What you do you think IT takes for no one to be able to do above eight five to keep increasing that number? You said like every increase in the number might require different yeah different improvements in the system yeah.

I think the nature of this focus in first the first answer is prony is I don't know this is you know edge of the research. So um again, nobody y's gotten to that number before. So what's next is gonna uh you know heroic a guess from my part um what we've been historically at different part of the step compile, next the different time points.

So I first in early, like three years ago or so, one of the major problems was just the latency of the little connection IT isn't just like the radio device wasn't super goo an earlier revision of the implant? And um I just like no matter how good your decoder was, if you think is updating every thirty milda seconds, fifty milda seconds, it's just gonna choppy. And um no matter how good you are, that's gonna frustrating and lead to chAllenges.

So you know at that point, IT was very clear that the main chAllenges just get the day off the device in a reliable way such that you can enable the next chAllenge to be to be tackled and then know at some point I was um you know actually the modeling chAllenge of how do you uh just build a good mapping like the supervised learning problem of you have a bunch of data and you have for a label, you try to predict just what is the right like neuro to architecture and hyper pym's to optimize. And that was a prom for a bit when you saw that he became a different panic. I think the next bond after that was actually just sort of software stability reliability um you know if you have widely varying sort of influence lencs um in your in your system or your you know you after lakes that every once in a while IT decreases your ability to making and get in a state of flow and IT based just disrupture control great.

And so there's a variety different software bugs and improvements were made to basically increase the performance of the system, made a much more reliable, much more stable and LED to us, a state where we could reliably lig data to build Better models with. So that was a about them for a while is just to like the soft stack itself. Um you've ought to guess right now, uh, there are sort of two major directions you could think about for improving B P.

S. further. Um the first major direction is labelling. So labelling is, again, this undammed tal chAllenge of given A A window of time with the users expressing some behavioral intent.

What are they really trying to do at the, at the granularity of every millisecond? And that, again, is a task design problem. It's A U. X problem. It's a machine learning problem, is to solve a problem, so touches all those different domains.

The second uh, thing you can think about to improve P P S further is either completely changing the thing you decoding or you extending the number of things you decoding. So this serve in the direction of functionality, they said. You imagine giving more clicks, for example, left to click or right click middle k uh, different actions like the condor g for example.

And that can improve the effectivity of your communication processes if you're trying to uh allow the user to express ourselves through any given location channel, you can measure that with this per second. But what actually measuring into the day is how effective are they and navigate the computer. And so from the perspective of the downstream task, you care about functionality, extending functionality. Something were very interested because not only to improve the number B, P, S, but he can also improve the downstream serve independence of user has on the skill efficiency with which they can Operate their computer.

Would the number of threads increasing also potentially help yes.

strangers, yes, to bid nuanced how that curve a or how that manifests in uh, in the numbers. So what you'll see is that if you apply uh curve of number of channels uh that you're using for code verse, either the offline metric of how good you are ready coding uh or the online metrics of of uh in practice how good is the user are using this device, you see roughly a law curve.

So as you move further out a number of channels, you get a cross burnings of log and improvement in control quality and offline validation metrics. The important new on here is that each channel corresponds with a specific, uh, you know, represented intention in the brain. So for example, if you have a channel two hundred fifty four and might crop with moving to the right, channel twenty fifty six might mean move to left if you want to expand the number functions you want to control, uh, you really want to have a broader set channels that covers a broader set of imagine movements.

If you think of IT like uh kind like mr. Potato and actually like if you had a point to different imagine moments you could do, how would you map those imagine movements to input to computer? Ah you could imagine you know, handwriting to output characters on the screen.

You could imagine just typing with your fingers and have that output text on the screen. You could imagine different fingering modulations for different clicks. You imagine wiggin s for opening some you or we in your know your big toe to have like, uh command taker or something like this. So it's really uh, the amount of different actions you can take in the world depends on how many channels you have an information content that they Carry.

right? So that's more about the number of action. So you actually is increased number of threats is more about increasing the number of actions able to perform one of the new one that worth mentioning.

So again, our goal is really to able to user with process to control the computer as fast as I can. That's B, P, S, uh, with all this in functionality I have so we just talked about but then also as a lively as I can yeah. And that last point is very related to channel count discussion. So has you scale out a number of channels the relative importance of any particular feature of your model input to the output control of the user diminishes, which means that if the set of neural mysa's ary effect is per channel, or if the nose is independent such that more channel means on average less output effect than your reliability, your system will improve. So one sort of course is um that is I have as a scaling channel count should improve the reliability and without any work on the counter itself.

can you link on the reliability here? So first of all, when you see stationary of the signal, which aspect are you reformed to? Yeah so baby.

let's talk briefly what the actual underlines and the looks like. So again, I I spoke very briefly at the beginning about how, when you imagine moving to the right or imagine moving to the left, neurons, my fire, more or less, and they are frequency content of that signal.

At least the motor cortex is very correlated with the output intention, the behavioral uh task with the users doing you could imagine actually this is not obvious rate coding, which is the name of that um phenomena is like the only way to make represent information. You imagine different ways in which the brain could in code uh intention and there's actually evidence like in bath for example, that there is temporal codes. So timing codes of like exactly win particular on fire is the the mechanism of information uh representation but at least the motor tex, the substantial evident that uh red coding, or at least one like first order factors that is coding.

So then if the brain is representing information by changing the so frequency of a non firing, what really matters is sort of the deal top between sort of the baseline said than are on and what IT looks like when it's modulated. And what we've observed, and what has also been observed in academic work, is that that baseline rate, if you tie the scale, if you imagine that analogy for like measuring, you know, flower something when you're baking that baseline state of how much the pot ways is actually different day to day. And so if what you trying to measure how much rises in the pot, you're going to get a different measures in different days because you're measuring with different parts.

So that based on rate shifting is really the thing that are at least from a first order description of the problem is was causing the downs am bias. There can be other effects now, any effects on top of that. But at least at a very first order description, the problem that we observed that days at the baseline firing of any particular neuron or observed on a particular .

channel changing. So can you just adjust to the baseline to make your relatives to the baseline stop?

Yeah this is a great question. Um with monkeys we have found various ways to do this. Um one example to do this is you ask them to do some behavioral task like play the game with a joystick.

You measure what's going on in the brain, you compute some mean of what's going on across all the input features, and you contract that at the input when you're doing your B, C, I session. Works super well forever. Reason that doesn't work super well with no land. I actually don't know the full reason why, but I can imagine several, several explanations. One such explanation could be that the context effect difference between some open new tasks task is much more significant with um no one that is with monkey.

Maybe in this open the task is you watching the three man podcast while he's doing the task or he's whistling and missing the music and talking with his friend and asked his more what's for dinner while he's doing this task? And so the exact sort of difference in context between those two states may be much larger. And that need to a bigger true journied ation gap between the features that you're are Normal alizon at open loop time and what you trying to use a close loop .

time that's interesting just on that point is kind of incredible to watch no. And be able to do to multitask, to do multiple task at the same time, to be able to move the mouse course are effectively while talking and while being nervous because he's .

talking us and chest to kicking .

your ass and now weren't talk trash while doing, yes, so all of the same time. And yes, if you are trying to Normalize to the baseline, that might throw everything off. Boy, is that interesting? Maybe one common on that.

too, for focus aren't familiar with a city technology. I think there's a common belief that what can you just use an attract or something like this, hoping somebody move a miles on the screen. And it's it's a really a fair question I won that I actually did was not confident before and that this was gonna a profound dly transformative technology for people like him.

And i'm very confident now that I will be. But the reasons are subtle that really have to do IT organically hypes into their life, even if you can just offer the same level of control as what they would have with an attacker or with a music. But you don't need to have that thing in your face.

You don't need a position a certain way. You don't need your caretaker to be around to set IT up for you. You can activate when you want, how you want, whatever you want, that love.

Independence is so game changing for people. IT means that they context the front night privately, without their moment to be in the loop. That means that dayan like, open up, you know, and browse the internet at two A M.

When nobody is around to their ipad up for them, this is like founded, game changing thing for folks in that situation. And this is even before we start talking about, folks at you may not be able to communicate at all or ask for help when they want to. This can be the potentially the only link that they have to the outside world. And yeah, that one doesn't, I think, need explanation .

of why that's so impatient. You mentioned neural decoder, how much machine learning is in the decoder, how much magic, how much science, how much art, how difficult is IT to come up? The decoder that figures out what is a sequence of Spikes mean?

A good question. Uh, there's a couple of ways to enter this. So you out briefly first and then i'll go down one of the rabbit holes.

And so to zone out, view that building the decoder, really the process to build the data set plus compiled into the weights and um you know those steps is important. Uh, the direction I think a further improvement is primarily going to be in the data. How do you construct the optimal levels for the model, but there's an entirely separate chAllenge of that, how you compiled the best model.

And so i'll go briefly down the second one, down the second riber hall. One of the main chAllenges with designing the optimal model for B, C, I is that alpine metrics don't aster the correspond to online metrics. Uh IT.

Finally, to the control problem, the user is trying to control something on the screen and the exact sort of user experience of how you output the intention, uh, impacts ability control. So for example, if you just look at validation loss as predicted by your model, there can be more ways to do the same validation loss. Not all of them are equally controlled by the end user s 的。

Now you might be a simple saying, oh, you add terms that, like, help you capture the thing that actually matters. But this is a very complex nuances process. So how you turn the labels into the model is uh more than nuances process than just like a standard to provide starting problem.

One very fascinating uh, anecdo. Here we've tried many different to nearly ork architectures that translate brain da to uh for osa output, for example. And one uh example that stuck in my brain from a couple years ago now, uh, is we at one point we are using just four kind of networks to decode the brain activity.

We tried uh A B task where we were measuring the realty performance in online control session of uh one d convolution over the inputs or so. If you imagine per channel, you have a sliding window that's producing some uh, compel future for each of those input sequences for every single channel simultaneously, you can actually get Better validation metrics, meaning you're fitting the data Better and it's journalizing Better. An offline data.

If you use this convolution architecture, you reducing premiers at sort of a standard, a stand, a procedure when to do them time, serious data. Now IT turns out that we're using that model online. The control of ability was was worse, was far worse even though the alpine metric around. And uh, there can be many ways to pret that.

But what that told me at least was, hey, it's at least the case right now that if you were to just throw a bunch of computer, this problem and you are trying to serve high primary optimize or you know, let some GPT model hard code or come up with your invent many different solutions, if you are just open for loss, IT would not be sufficient, which means that there's still some inherent model in gap here. So some ortis ry left to be uncovered here, how to get your model to scale with more compute. And that may be fundamentally labelling problem, but there may be other companies as well.

Is IT a data constrained this time? Like the which is what IT sounds like, like how do you get a lot of good labels?

Yeah I think it's date a quality constraint.

not also did a quality constraints, but even like good, even just the quantity. I mean, because he has to be trained at the on the interactions. I guess there's not that many interactions.

Yeah so IT depends what version is you talking about. So I like but the simplest example of just two famos ity, then I think you did quality is aid them if you're talking about how to build a sort of multi functions output that let you do all the inputs, the computer that you and I can do, then it's actually a much more sophisticated new once model in chAllenge.

Because now you need to think about not just when the users left locking, but when you're build the left like model, you also need to be thinking about how to make sure a dozen and fire when they trying to right like, or they trying to move the most. So one example of an interesting bug from, like sort of week one, A, B, C, I. No one was when he moved the mouse uh, the clicks have dropped off a Cliff.

N when he start the clicks and went up. So there's contamination between the two, the two inputs and all. The good example was one point he was trying to do, uh, sort of a laugh, looking drag, and the minute he started moving, the left click signal drop off a Cliff. So again, because of some contamination between two signals, you need to come up with some way to either in the dataset or in the model, build robust against this kind of, uh, you think they like over fitting, but really it's just that the model has not seen this kind of very ability before, so you need to find some way to help the model with.

This is very cool. I think he feels like all of this is very solvable, but it's hard. Yes, IT is fundamentally .

engineering change. This is important, emphasize and is also important, emphasize that I may not need funding in new technique, which means that you know, people who work on, let's say, how to revise speech classification using C, T, C, loss, for example, with internal theory, they could potentially a very applicable skills to this.

So what things are you excited about in the future development? Have the software stack on on your link? So everything we had been talking about, the decode in the U. X.

I think there's someone excited about like something about from the technology side and someone excited about understanding how this technology is going to be based in front in the world. So our backwards on the technology entering the world's of things, i'm really, really excited to understand how this device works for folks that uh you know cannot speak at all, that have no ability to sort boot, struct myself into useful control by voice command, for example, and are extremely limited in the current capabilities. I think that will be incredible.

Useful enough to understand, I mean, really, what is an actual time for all startups, which is part of market fit? Does this device have the capacity and potential to transform people's lives in the current state? And if not, what are the gaps? And if we are, are gaps, how do we solve the most, most efficiently? So that's what i'm very excited about for the next years of clinical trial Operations.

The um technology side, uh, i'm quite exact about basically everything we're doing. I think, uh, it's gonna be awesome. The most prominent one out IT is, is scaling channel count.

So right now we have a thousand channel device. The next version will have between three and six thousand channels and expect that curve to continue in the future. And it's unclear what set of problems will just disappear completely at that scale and what site problems will remain in require for their focus.

And so i'm excited about the clarity of gradient that, that gives us in terms of the use experiences that we choose to focus our time and resources on and IT also in terms of the even things are most noticeable like does, I promise, completely go away that scale? Or do you need to come up with no creative you access still even at that point um and also want to get to that time point when we start expanding automatically the set of functions that you can output from one rain. How to deal with all the nuances of both the user experience of not being able to feel the different keys under your fingertips, but still need to build to module all of them.

And so you need to achieve that thing you want. And you can you don't have that proper set to feed back loop. So how can you make that intuitive for user control, a high dimensional control surface without feeling the thing physically? I think that's going a super interesting problem.

I'm also quite excited to understand, you know, do these scaling laws continue like as you scale channel count, how much further out do you go before that situation? Point is is truly head and so obvious today. I think we only know what's in the of interpretation space.

We know what's between zero one thousand twenty four. We don't know what on that. Um and there is a whole set of like range of interesting sort of neuroscience and brain questions, which is when you stick more stuff in the brain in more places, you get to learn much more quick about what those brain regions represent. And so i'm excited about that fundamental of news science learning, which is also important for now how to most efficiently insert of a child in the future. So yeah, I think all those matters i'm really, really excited about that is even get close to touching this of software stack that we work on every day we were working on right now.

Yeah, IT seems virtually impossible to me that A A thousand electrodes where saturates IT feels like this would be one of those silly notions in the future where obviously should have millions of electrodes. And this is where, like, the true breakthrough happened. Yeah, you tweet. Now, some thoughts are most precisely described in poetry. What do you think that is?

I think is because the the information bottle mackle language is a pretty steped and yet you're like you're able to reconstruct on the other persons and the other person's brain more effectively without being literal like if you if you can express the sentiment such in their brain, they can reconstruct the the actual true underlying meaning and beauty of the thing that you're trying to get across the the generator functioning their brains more powerful than what language can express. And so uh the mechanism poetry is really just to uh feed or seed that generator function being literal .

sometimes is a sub optimal compression for for the thing you trying to convey.

And it's actually in the process of the user going to that generation that they understand what what you mean. Like that's the that's the beautiful part is also like when you look at a beautiful painting, it's not the the pixel of the painting that are beautiful. It's the job process that occurs when you see that the the experience of that after that matters.

He has rested with some .

deep you thing within .

you that the artists also experienced that was able to convey that through ics. And that's actually going to be relevant for for fall on telepathy, you know is is like if you just read the poetry literally, that doesn't say much of anything interesting. They requires a human to interpret IT.

So it's the combination of the human mind and all the experiences the human being has within the context of the collective intelligence of the human species that makes that power makes sense. And they load that in. And so in that same way, the signal that Carries from human to human meaning might not, may seem trivial, but may actually care a lot of power because of the complexity of the human mind on the receiving end.

Yeah, that's interesting. I post. He still doesn't. Who's IT? I think, uh, you should about a first choice at something about all the people that think we've achieved A G I explain why humans like music oh yeah and and until until the I likes music, you achieved A G I something like.

do I think that's like some next token entrepreneur's SE I thing going? I don't know, I don't know either. I listened to a lot of classical music and also read a lot of poetry. And yeah, I do wonder if, like there is some problem of the next token surprise fact going on there. Yeah, maybe they like a lot of the the tricks.

And both poetry and musics are like basically have some repeat the structure and then you do like a twist ah like like okay first or like claws two three is one thing and then close for like okay now we're under the next team yeah and they kind of play with exactly when the surprise happens and expectation of the user. And that's even true. Like through history, as musicians evolve music, they take like some known structure that people are familiar, and they just tweet IT a little bit like a tweet and add a surprising element. This is especially true like in classical music erith, that's one is all just enrobed.

Like break of breaking structure or breaking symmetry is something that humans seem to like.

maybe as simple as that. yeah. I mean, great. Are this copy ah and they also know knowing which rules to break as the important part. And IT financially IT must be about the the listener of the piece. Like which rules is the right one to break is about the user, the audience, of perceiving that as interesting.

Uh, what do you think is the meaning of human existence?

There's A T V show I really like called the west wing and in in the west wing there's a charactery, the president of united states whose are having discussion about the bible with one of their colleagues and uh what the colleague said something about, you know, the bibles as X, Y, N, Z and uh, the present says, A, B, C, and person says, too well, do you believe the bio to be literally true and the present says.

yes. But I also think .

that neither of us a song enough to understand, I think to like technology here for the meaning of life is that largely we don't know the right question to ask. And so I think i'm very like with um sorry, the hitchhikers guide, the galaxy version of this question, which is basically if we can ask the right questions, it's much more likely we find the meaning of human existence and so in the short time is a hurry tic in the sort of search policy space. We should try to creed diversity of a people asking such questions or generally of conscious ness and conscious beings asking such questions. Um so again, I think I will take the I don't know card here but said I do think there are meaningful things we can do that improve a likely to enter question.

It's interesting how much value you assigned to the task of asking the right questions. That's the that's the main thing is nothing answers to the questions this point.

by the way, driven home are in a very painful way when you try to community with someone who cannot speak. Because a lot of the time, the last thing to go is they have the ability to somehow, you know, with the a lip or move something that allows them to say yes or no. And in that situation, it's very obvious that what matters are you asking them the right question to be so yes or .

no to wall as powerful? Well, plus, thank you for everything you do. And thank you for being you and thank you for talking to.

Thank you. Thanks for listening to this conversation with bliss chapman. And now their friends has known arba the first human being to have a newly link device planted in his brain. You had a diving accident in twenty sixteen that left you paralyze with no feeling from the shoulders down. How did that accident change your life?

This is sort of a freak thing that happened imagining your running into the ocean. Um this doesn't like but you're running into the ocean and you get to about waste high and then you kind of like dive in, take the rest of the plunge under the way or something. That's what I did.

Um and then I just never came back at not sure what happened. Uh, I did IT running into the water with a couple guys. And so my idea of what happened is really just that. I took like a stray fist, elbow, nee foot, something to the side of my head. The lift start of my head was sore for about a month afterward, so must be taken a pretty big knock and then they both came up and I didn't and so I was faced down on the water for a while I was conscious um and then eventually just you know realized I couldn't all my breath and any longer and I keep saying took a big drink um people I don't know, they like that I say that seems like i'm making light at all but just kind of how I am and I don't know like i'm a very relaxed sort of stress free person. I rolled with the punches for a lot of this.

I kind of took IT in stride like, alright, well, what can I do next? How can I improve my life even a little bit on a data day basis at first? Just trying to find some way to heal as much my body as possible um to try to get healed, to try to get off a ventilator um learn as much as I could so I could somehow survive um once I left the hospital um and then thank god I had like my family around me if I didn't have my parents my siblings and I would never met this far。 They've done so much for me more than like I can ever thank them for honestly.

And a lot of people don't have that. A lot of people in my situation, their families either aren't capable of providing for them or honestly just don't want to. And so they get place somewhere, you know in some sort of home.

Ah so thankfully, I had my family. I have great group of friends, a great group of bodies from college who have all riled around me. And we're all i'm still incredibly close.

People always say, you know, if you're lucky, you'll end up with one or two friends from high school that you keep without your life. I have about ten, ten or twelve from high school that have all stuck around. And we still get together all of us twice a year.

We call IT the spring series and the fall series um this last one we all did. We dressed up like x men. So I did the best acai or and IT was frequent awesome.

IT was so good. So yeah, I have such a great supporters around me. And so you know, being a quarter politica isn't that bad. I get waited on um all the time.

People bring me food and drinks and I get to sit around and watch as much T, V and movies and enemy as I won. I get to read as much as I want. I mean, it's it's great.

It's beautiful to see that you see the silver lining in all this. We just going back. Do you member, at the moment when you first realize you are paralyzed from a nighttime?

Yep, I was face down in the water, right? When I, whatever something had my head, I tried to get up and I realized I could move and it's just sort of click like I I am paralyzed, can't move. What do I do um I can't get up, I can't flip over, can do anything then i'm gonna drown eventually um and I knew I couldn't hold my breath forever so I just held my breath.

Thought about IT for maybe ten fifteen seconds. Um i've heard from other people that like on lookers, I guess the two girls that pulled me out of the water were two my best friends. They are lifeguards um and one of them said that um IT looked like my body was sort of shaking in the water like I was trying to flip over and self um but I knew I knew immediately and I just kind of I realized that that's like what my situation was from here on out.

Maybe if I got to the hospital to be able to do something when I was in the hospital like ripe for surgery, I was trying to calm a one of my friends down. I'd like brought her with me from college to camp and he was just bowling over me and as a kay, be fine. Like don't worry. Um I was cracking some jokes to try to light in the mood.

Um the nurse had called my mom and I don't my mom, she's just going to be stressed out, called her after him at a surgery because at least i'll have some answers then like whether I live or not really and I didn't want her to be stressed through the whole thing but I knew and then when I first woke up after surgery, um I was super drugged up they had me on fatal like three ways IT was awesome. Um I don't I don't recommend IT but um I saw I saw some crazy stuff uh on that fit all and IT was still the best i've ever felt on drugs medication side on medication um and I remember the first time I saw my mom in the hospital, I was just bowling. I had like ventilator in um like I could not talk to anything and uh I just started crying because I was more like seeing her not that I mean the whole situation obviously was pretty rough but I was just like seeing her face for the first time was pretty hard but um yeah I I never had like a moment of, you know man, this sucks. I don't want to like me around anymore IT was always just I hate that I have to do this but like sitting here in willows ing isn't going to help.

So immediate acceptance .

yeah yeah.

Has there been low points long way?

Yeah yeah 是 嗯， i mean, there are days when I don't really feel like doing anything not so much anymore. Like not for the last couple years. I don't really feel that way.

I've more so just wanted to try to do anything possible to make my life Better at this point. Um but at the beginning, there are some up and doing. There are some really hard things to adjust to.

Um first off, just like the first couple months, the a amount of pain I was in was really, really hard. I mean, I remember screaming at the top of my lungs in the hospital because I thought my legs run fire and obviously I can't feel anything, but it's all nerve pain. And so that was a really hard night.

I asked them to give me as much pain that is possible, though, like you ve had as much as you can have. So just kind of deal with IT, go to a happy place sort of thing. So that was a pretty low point.

Um and then every now again, it's hard like realizing things that I wanted to do my life, that I won't be able to doing more um you know I always wanted to be a husband and father and I just don't think that I could do IT now as a quality project maybe it's possible but i'm not sure I whatever um put you know someone I love through that um like having to take care of mean stuff um not being able to you know go out and play sport. I was a he was athlete ally growing up. I was pretty hard um little things do when I realized I can do them anymore.

Like there's something really special about being able to hold the book and smell L A book like the feel, uh, the texture, the smell like as you turn the pages like I just love IT and I can't do anymore and it's little things like that. Um the two year mark was pretty rough two years when they say you will get back basically as much as you're ever gonna back as far as movement in sensation goes. And so for for the first two years that was the only thing on my mind was like try as much as I can to move my fingers, my hands, my feet, everything possible to try to get sensation and movement back.

And then when the two year market, so um june thirty, twenty eighteen, I was I was really sad that that kind of where I was um and I just randomly here there but I was never like the Price for long periods of time. Just IT never seemed worthwhile to me. But gave his strength my faith, my facing god was a big one.

My understanding that I was all for purpose even if that purpose wasn't anything involving nearly, even if that purpose was, you know, there's there's a story in the bible about job. And I think it's a really, really popular story about how jobe, you know, has all of these terrible things happen to many Prices got throughout the whole situation. I thought, and I think a lot of people think for most of their lives that they are job, that they are the ones going through something terrible, and they just need to, you know, praise god through the whole thing, and everything will work out.

At some point. After my accident, I realized that I might not be job, that I might be, you know, one of his children that gets killed or kidnap or taken from him. And so it's about terrible things that happened to those around you who you love.

So maybe, you know, in this case, my mom would be job, and he has to get through something extraordinary hard. And I just need to try and make IT as best as possible for her because um she's the one that really going through this massive trial um and that gave me a lot of strength. And obviously my family um my family, my friends, they they give me all the strength that I need on a da devis, so makes things a lot easier. You're having that great support system where .

on me from everything i've seen of you online, your dreams and the way you are today, I really admire, let's say you're on waivers, positive outlook on life. Has that always been this way?

Yeah, yeah. I I mean, I just always thought I could do anything I ever wanted to do. There was never anything too big, like whatever I set my mind too. I fall like I could do IT. Um I didn't want to do a lot.

I wanted to like travel around and be sort of like a gypsy, and like go work odd jobs at the stream of traveling around european, being like, I don't know, a shepard in like whales or ireland and then going being a Fishermen in italy doing all these things like a year. I get such like cliche things, but I just thought I would be so much fun to go in travel and do different things. And so um i've always just seen the best in people around me too and i've always tried to be good to people and grown up with my mom too.

She's like the most positive, energetic person in the world and were just people people like, uh I just get along great with people. I really enjoy meeting new people and so um I just wanted to do everything. Um this is kind of just how i've been IT is great .

to see that cynisca not take over given everything you have been through yeah was like a deliberate choice made you are gonna let .

this keep IT down yeah a bit also like I just just kind of how I am. I just like I said, I roll the punches of everything I was used to tell people like I don't stress about things much um and whenever i'd see people getting stressed or just say, you know like it's not hard, just don't stress about IT and like that's all you need to do uh and they're like that's not how that works like that works for me.

I just don't stress and everything will be fine. Like everything work out. Obviously not everything always goes well and not like IT all works out for the best of the time, but I just don't think stress has any place in my life since I was a kid.

Who is the experience, like, of you being selected to be the first human being to have annual device planting your brain? We scared. I said.

no, no. I was .

school like I was.

I was never afraid of IT. I had to think through a lot. Should I should I do this um like be the first person I could wait till number two or three and get a Better version of the newer link like the first one might not work maybe um it's actually gonna can't suck.

It's going to be the worst version ever in a person. So why would I do the first one like I art kind of selected. I could just tell them, you know like okay, find someone else and then i'll do number two or three like i'm sure they would let me.

They're looking for a few people anyways but ultimately I was like, I don't know there's something about being the first one to do something. It's pretty cool. I always thought that if I had the chance that I would like to do something for the first time um this seems like a pretty good opportunity and I was I was never scared.

I think I like faith had a huge a part that I was felt like god was preparing me for something um I almost wish I wasn't this because I had many conversations with got about not wanting to do any of this as a great political I told them, you know, i'll go out and talk to people, go and travel the world and talk to new stadium. Thousands of people give my testing money. I'll do all of that.

But like, kill me first. Don't make me do all this in a chair that sucks. Um and I guess he won that argument. I didn't really have much of a choice. I was felt like there was something going on. And to see how I get easily, I made IT through the interview process and how quickly everything happened um how the star sort of a line to solve this is IT just told me like as the surgery was getting closer, IT just told me that you know IT IT was all meant to happen, was all meant to be and so I shouldn't afraid of anything that to come and so I wasn't I kept telling myself like, you know, you say that now.

But as soon as the surgery comes, you're probably going to be freaking out like you about to have brain surgery and brain surgery is a big deal for a lot of people, but it's a even bigger deal for me like it's all I have left the amount at times I ve been like thank you god that you didn't even take my brain in my personality, in my ability to think um my like love of learning, like my character, everything like thank you so much like as long you left me that then I think I can get by and I was about to let people go like world around and there like we're going to go like put some stuff in your brain like hopefully IT works out um and so I was IT was something that gave me pause but like I said, how smoothly everything went, I never expected for a second that anything could go wrong. Plus the more people I met on the bar side and on the nearing side, they're just the most impressive people in the world. Like, I can speak enough to how much I trust these people with my life and how impressed I am with all them, and to see the excitement on their faces.

So like, walk into a room and roll into a room and see all of these people looking at me, like we're just we're so excited, like we've working so hard on this and it's finally happening. It's super infectious. And um IT just makes me want to do IT even more and help them achieve their dreams like you to know it's so it's overwarm and i'm so happy for all of them honestly.

was the day of surgery like what's when you wake up what you feel yeah minute by minute yeah we you .

free out no, I thought I was gone to but the surgery approach the night before the morning of I was just excited I I like, let's make this happen I think I said that, uh, something like that iron on the phone beforehand. We were like a face timing and I was like, let's rock and roll and he's like, let's do IT. Uh, I I don't know.

I just, I wasn't scared. So we woke up. I think we had to be at the hospital like five thirty A M.

I think surgery was seven. So we woke up pretty early. I'm not sure much of us slept that night. Got to the hospital, five thirty went through like all the pro p stuff, everyone was super nice. Uh elan was supposed to be there in the morning um but something went wrong with his plane.

So we need the face timing that was school had one of the greatest one liners of my life after that phone call um hung g up with them. They were like twenty people around me. Now I god just hope he wasn't too star struck talking to me. Yeah, good. All done.

Yeah yeah. Time now came to me.

I was like, this is this seems right you know, when in the surgery, um I asked if I could pray right beer hands so I like prae over the room I ask god if you like me with my mom in case anything happened to me and I just like calm her nerves out there. Uh, woke up and played a bit of a prank on my mom. You've heard about IT yeah, I read about yeah .

he he was not happy.

Can you take me to the prank?

Yeah this is regret .

doing that now no um IT was something that was something I I had talked about ahead of time with my buddy bin. I was like I would really like to play a print on my mom um very specifically my mom, she's very golden. I think he had me surgery once even and um after he came out in the surgery um uh SHE was super groggy SHE was like I can feel my legs and my dad look that just like you don't have any legs like they had, they had to amputate both your legs and we just do very mean things to her all the time um i'm so surprised that he still loves us um but right after surgery I was really worried that I was going to be too like groggy like not all there I had had an a once before and IT IT messed me up like I could not function for a while afterwards and I um I like said a lot of things that I was like I was really worried that I was gonna start, I don't know, like drop in drop in some bombs and I wouldn't even know.

I wouldn't remember um so I was like like, please guy, I don't let that happen and please let me be there enough to do this to my mom and so SHE walked in a after surgery was like the first time we'd been able to see me after surgery and SHE just looked to me, she's hi, like, how are you? How are you doing? How do you feel? And I looked at this very, I think the everything I hope very like gi sort confused look on my face, who are you? And SHE just started looking around the room like at the surgeons of the doctors.

Like, what did you do to my son? Like, you need to fix this right now? Tears started dreaming.

I saw how much he was freaking out. I was like, I can't let this go on one fine. Like, uh, it's alright.

And, uh, still he was not happy about IT. SHE ah still says she's going to get me back someday. But I mean, I I don't know, I don't know. That's gonna like .

a bong battle. Yeah yeah. But IT was good in some sense. That was a demonstration that .

he still got all I wanted, all I wanted to be. And I knew that doing something super mean to her like that would show her. So yeah.

to show that you still there.

that you love yeah, exactly.

exactly the dark way to do IT. But I love IT. Yeah, what was the first time you're able to feel that you can use the new link device to affect the world around you?

Yeah um the first little taste I got of IT was a actually not too long after surgery um some of the nearing team had brought in um like a little ipad um a little tablet screen and theyd put up eight different channels um that we're recording some minor on Spikes um when they put IT in front of me like this is like real time your brain firing like that super cool um my first thought was, I mean, if they're firing now, lets see if I can affect them in some way so I started trying to like we go my fingers and I just started like scanning through the channels.

And one of the things I was doing was like moving my index finger up and down. And I just saw this yellow Spike. I'm like top o like third box over or something.

I saw this yellow Spike every time. I did IT not like, that's cool. And everyone around just like what what you seen like, look, look at this one.

Look at like this top row third box over this year. bike. Like, that's me right there, there there.

And everyone was freaking in out. They started like clapping and like that super unnecessary. Like, this is what's supposed to happen.

right? Like, so you're imagining yourself moving each individual feature one at a time, yes. And then seeing like they did the index finger, you're like home .

yeah I I was wiggling kind of all of my fingers to see if anything would happen. There was a lot of other things going on, but that big yellow Spike was the one that stood out to me like, i'm sure that if I would stare at IT long enough, I could have mapped out maybe one hundred different things. But the big yal Spike was the one that I noticed.

Maybe you can speak to what is like to sort of wiggle your fingers, that like to to imagine that the mental, the coding of effort required to sort of wiggle your index finger, for example, how easy that to do.

Pretty easy for me. It's something that at the very beginning, after my accident, they told me to try and move my body as much as possible even if you know you can just keep trying because that's going to create new like neural pathways or pathways in my spinal cord to like reconnect these things um to hopefully regain some movement someday yeah I know it's bizarre, but is part of .

the recovery processes to keep trying to move your body and and and the nervous system does this thing yeah reject.

It'll start reconnecting for some people. Some people never works. Some people they'll do IT like for me, I got some bye p control back um and that's about IT. I can if I try enough, I can wigged some of my fingers not like on command. It's more like if I trying to move, say my right pinky and I just keep trying to move IT after a few seconds at a wiggle um so I know their stuff there, like I know like that happens with you know a few different of my fingers and stuff. Um but yeah that's that's what they tell you to do.

Um one of the people at the time when I was the hospital came in and told me for one guy who had recovered um most of his control, what he thought about everyday was actually walking like the act of walking just over and over again. So I tried that for years. I tried just imagining walking, which is it's hard it's hard to imagine like all of the steps that go into well taking a step, like all of the things that have to move, like all the activations, uh, that have to happen along your leg in order for one step to occur.

But you're not just imagining you're like doing IT.

I'm trying I it's like it's imagining over again what I had to do to take a step. This is not something any of us think about. We just you want to walk and you take a step.

Um you don't think about all of the different things that are going on in your body. So I had to recreate that in my head as much as I could and then I practice IT over and over. And ops is now .

I get a third person perspective as a first person perspective. You like it's not like you're imagining yourself walking. You're like literally doing is everything all the same self as .

if you're walking which which was hard IT was hard at .

the beginning like frustrating ing hard or like actually cognate .

hard like IT was both um. There's there's a theme in one of the king bill movies actually, uh oddly enough, where he is like paralysed, I don't know, from like a drug that was in the system and then SHE like finds some way to get into the back of a truck or something. And SHE stares at her toe and he says, move, like, move your big toe.

And uh, after you know, a few seconds on screen, SHE does IT. And he did that with every one of her like body parts until he can move. I did that for years, just stared at my body instead, move your index finger, move your big toe.

Um sometimes vocalizing IT like out loud, sometimes just thinking IT. I tried every different way to do this to try to get some movement back. And it's hard because IT actually is like taxing like physically taxing on my body, which is something I would have never expect because it's not like i'm moving. But IT feels like there's a build up of know the the only way I can describe .

IT .

is they are like signals that are getting through from my brain um down because my there's a gap, my sports of brain down and then from my hand back up to the brain. And so IT feels like those signals um gets stuck in whatever body parts that i'm trying to move. And they just build up and build up and build up until they burst.

Um and then once they burst, I get like this really weird sensation of everything sort of like dissipating back out to level and then I do IT again. Um it's also just like a fatigue thing, like a muscle fatigue, but without actually moving your muscles, it's very, very barre. And then you know if you try to stare at a body part or think about a body part and move for to three, four, sometimes eight hours, it's very taxing on your mind.

It's takes a lot of focus. Um IT was a lot easier at the beginning because I wasn't able to like control A T V in my room or anything. I was able to um control any of my environment. So for the first few years, a lot of what I was doing was staring at walls. And so um obviously I did a lot of thinking and I try to move a lot just over and over and over again.

Do you never give out of hope there? Training hard.

essential here. And I still do IT. I do IT likes so consciously. And I think that a that helped a lot with things with neurolinguistic sty, something that I talked about the other day at the all .

hands and .

I did at nurlan Austin facility. welcome. And yeah, hey, the gig of factory was super cool. I went to school at text in. So i've been around .

for see you say welcome to me yes, welcome to text.

But I was talking about how a lot of what they've had me do, especially at the beginning, what I still do that now um is body mapping. So like there will be a visualization of a hand or an ARM on the screen and I have to do that motion and the tidy sort of train um the algorithm to like understand what i'm trying to do. And so IT made things very same as for me. I think that's really.

really cool. So it's it's amazing to know because I i've learned a lot about the body mapping procedure yeah I like with with the interface and like that is cool to know that a century like training to .

be like world class at that test yeah yeah I I don't know if other quality gic, like other paralyzed people give up I hope they don't um I hope they keep trying because i've heard other paralyzed people say like don't never stop they tell you two years but um you you just never know you you the human bodies capable of amazing things so um i've heard other people say don't give up uh like I think one girl had um spoke to me through some family members and said that he had been paralyze you know for eighteen years and sh'd been trying to like wigg her index finger for all that time and SHE finally got a bat like eighteen years later so like I know that is possible and i'll never give up doing IT.

I just I do IT when i'm lying down like watching T V, i'll find myself doing IT can I just almost like on IT don't this is something i've gotten so used to doing that? I don't know. Like I don't think I think it's really awesome .

to hear because I think it's one of those things that can really pay off. And and because like that is training. You're not visibly seen there is out of that training at the moment. But like there's like like olympic level .

nervous something that I think na link gave me that um I can I can't think them enough for like I can't show my appreciation for enough was being able to visually see that what i'm doing is actually having some effect yeah um it's a huge part of the reason why like I know now that i'm gonna keep doing IT forever because before nearly I was doing IT every day and I was just assuming that things were happening like it's not like I knew I wasn't getting back any mobility or um sensation or anything so I could have been running up against a brick wall for all I knew and with no link I get to see like all the signals happening real time and I get to see that you know what i'm doing can actually be mapped.

You know, when we started doing like click calibrations and stuff, when I go to click my index finger for a left click that had actually recognizes that like IT IT changed how I think about what's possible with like bring my body to move. And so i'll i'll never give up. And also just a signal .

that there is still a powerhouse of a brain air. That's like I and as the technology developed, that brain is, I mean, that's the most important thing about the human bodies, the brain. You can do a lot of the control. So what did you feel like when you first could wigg the index finger and saw the environment respond like that? Yeah think whatever we just being way to dramatic .

according to yeah, IT was very cool. I mean, IT was cool. But IT, I keep telling us to people and made sense to me, like I made sense that, you know, like there are signals still happening in my brain, and that as long as you had something near IT that could measure those, that could record those, then you should be able to like visualize IT in some way, like see that happen.

And so that was not very surprising to me. I was like, oh cool. Like we we found one, like we found something that works.

Um IT was cool to see that their technology worked um and then everything that theyd worked so hard for was like going to pay off um but I don't like moved, curse or anything at that point. I had like interact with a computer or anything at that point. Um so IT IT IT just made that that was cool. Like I I didn't really know much about B C I at that point either. So I didn't know like what sort of step this was actually making. Um like I didn't know if this was like a huge deal or if this was just like, okay, this is you know it's cool that we got this far, but we're actually hoping for something like much Better down the road like, okay, I just thought that they knew that IT turned on so I was like a what you like .

read up on the spects of the hard way. Get installed like the number threads yeah yeah.

I know all that, but it's all like, so greek to me. I was like, okay, threads, sixty four threads, teen electrics, thousand twenty four channels. Okay, like, like that that math checks out .

sounds right. yeah. What was the first time you were able to move a mouse course?

I know that must have been within the first, maybe weak week or two weeks that I was able to, like, first move the curse. And again, like IT kind of made sense to me. Like IT IT didn't seem like that big of deal. Like IT IT was like, okay, well, how do I explain this when everyone around you started clamping for something dan, it's it's easy to say, okay like I did something cool like that was that was impressive in some way um what exactly that meant what IT was hadn't really like set in for me um so again, I knew that me trying to move a body part and then that being mapped in some sort of like machine learning algorithm to be able to um identify like my brain signals and then take that and give me current control that all kind of sense me I don't know like all the interview IT. I was like there are so signals in my brain firing they just can't get through because there's like a gap in my spare cord and so they just they can cut all down and back up but they're still there so when I move the cursor for the first time was like that's cool but I expected that that should happen like IT IT made sense to me um when I moved the curse for the first time um with just my mind without like physically trying to move so I guess I can get into that just a little bit like the difference tween attempted movement and imagined movement .

yeah that's the first difference yeah the other .

yeah yeah yeah. So like attempted movement is me physically trying to attempt to move, say, my hand I try to attempt to move my hand to the right, to the left, foreign back um and that's all attempt attempt to you know like lift my finger up and down, attempt to kick or something um i'm physically trying to do all of those things even if you can't see you again. This would be like me attempting to like shrug my shoulders or something that's all attempted movement um that all that's what I was doing for the first couple of weeks when they were going to give me curse control.

When I was doing body mapping IT was attempt to do this, attempt to do that when um near was telling me um to like imagine doing IT, it's like kind of made sense to me but it's not something that people practice like if you started school um as a child and they said OK write your name with this pencil and so you do that like okay, now imagine writing your name with that pencil kids would think, uh like I guess like that kind of makes sense and they would do IT um but that's not something we're taught. It's all like how to do things physically. We think about like thought experiments and things.

But that's not like, that's not like a physical action of doing things is more like what you would do in certain situations. So imagine movement IT IT never really connected with me, like I guess you could. Maybe described this as, like a professional athlete, like has swing a baseball, but swing like a golf club.

Like imagine what you're supposed to do, but then you go right to that and physically do IT like you. Then you get a bat in your hand and then you do what you've been imagining and so I don't have that like connection. So telling me to imagine something versus attempting IT IT just there wasn't a lot that I could do there. Um mentally I just kind of have to accept what was going on and try um but the attempt moving thing IT all made sense to me like if I try to move then there's a signal being sent in my brain and as lunas they can pick that up, then they should be able to map IT to what i'm trying to do and so when I first moved the curse like that IT was IT was like, yes, this should happen like i'm i'm not surprised by that. What can you clarify?

Is this supposed to be a difference? Imagine movement and attempted movement .

yeah just that in imagine movement you're not attempting to move at all.

So it's like visualize and then theoretically supposed to be a different part of the brain that lies up in those two different situations.

Yeah, not necessarily. I think all these things can still be represented in model cortex. But the difference, I think, has to do with the natural of god, and by the on .

the light is bliss. So like this is just different ways to prompt you to kind of get to the thing that you are. Yeah, attempted movement does sounds like the right thing. Yeah, try.

Yeah, I only makes sense .

to me because imagine for me i'll be i'll start visualizing like in my mind visualizing attempted I would actually start trying to like, yeah, there's I mean, I did not come as with my whole effec wrestling when i'm imagining a move. See, i'm like moving my muscle exactly like there is a bit of an activation almost verses like visualizing yourself like a picture doing IT yeah .

it's something that I feel like naturally anyone would do if you try to tell someone to imagine doing something, they might close their eyes and then start at physically doing IT um but just ick.

yeah it's it's hard .

IT was very hard at the beginning but attempted worked, attempted work. They work just like I should work like, worked like a charm.

There is like one tuesday we were messing around and I think I forget what's wear war you used but there is a swear that came out your mouth when you figured out you could just do the direct control.

Yes, that IT IT blew my mind like no one intended, blew my mind when I first um moved the curse just with my thoughts and not attempting to move. It's something that I found um like over a couple of weeks like building up to that um that as I get Better current controls like the model gets Better than IT gets easier for me to like um like I don't have to attempt is much to move in and part of that is something that i'd even talked with them about um when I was watching the signals of my brain one day, I was watching when I like attempted to move to the right and I watched the screen as like I saw the Spikes like I was seeing despite the signals being sent before I was actually attempting to move.

Um I imagine just because you know when you go to say move your hand or anybody part that signal get sent before you're actually moving has to make IT all the way down and back up before you actually do any sort of movement. So there's a delay there. And I noticed that there was something going on in my brain before I was actually attempting to move, that my brain was like anticipating what I wanted to do.

And that all started sort of um I don't know, like percuss ating in my brain. I get just use sort of there like always in the back like that so weird that I could do that is kind of makes sense but I wonder what that means as far as like using the neuroses link. And um you know then as I was playing around with the attempted movement and playing around the curse and I saw that like as the cursor control got Better, that IT was anticipating my movements, what I want to to do like current movements, what I want to to do a bit Better and a bit Better.

And then one day I just randomly is also playing what good I am, like looked at a target before I had started like attempting to move. I was just trying to like get over, like train um my eyes to start looking ahead. Look okay, this is the target i'm on.

But if I look over here to this target, I know I can like maybe a bit quicker getting there and I looked over and the Carter just shot over IT IT was a child. Like, I had to take a step back. I was like, this should not be happening all day.

I was just smiling us. So getting as I guys do you know that this works like, I can just think IT and IT happens, which like they have all been saying this entire time, like I can't believe like you're doing all this with your mind and like that but isn't really with my mind. Like i'm attempting to move and he just picking that up so he doesn't feel like it's with my mind.

But when I moved IT for the first time, like that, IT was, oh, man, you like, IT made me think that this technology, that what i'm doing, is actually we way more impressive than I ever thought. I was very cool than I ve thought. And IT just opened up a whole new world of possibilities of, like what could possibly happen with this technology and what I might be able to be capable of with IT.

Because you had felt for the first time, like this was digital telepathy, like you're controlling a digital device with your mind. I mean, this was a real moment of discovery that's really cool. Like you've discovered something i've seen like scientists talk about like a big aham omit you like nobel Price when theyll have this like, holy crap yeah like that's that's what I felt .

like I got didn't feel like like I feel like I had discovered something. But for me, maybe not necessarily for like the world at large or like this field at large, IT just felt like an a hut moment for me. Like, oh, this works.

Like, obviously IT works. And so that's what I do like all the time. Now I kind of intermix um the attempted movement and a imagine movement I do IT all like together because I found that there is some interplay with IT that maximizes efficiency with the cursor.

So it's not all like one or the other. It's not all just I only use attempted or I only use like imagine movements. It's more I use them in parallel and I can do one of the other.

I can just completely think about um whatever i'm doing. But um I don't know. I I like the playground. They also like to just experiment with these things, like every now again, i'll get this idea in my head like, I wonder if this works and I just start doing that.

And then afterwards I tell them, by the way, I wasn't doing that like you guys wanted me to. Um I was. I thought of something and I anted to try IT, and so I did IT seems like IT work. So maybe we should like, explore that a little bit.

So I think that discovery is not just for you, at least for my prospect. That's a discovery for everyone else. Whoever uses in your link that this is possible, like I don't think this an obvious thing that this is even possible.

It's like i'll saying to bliss earlier, I like the four minute mile people thought was impossible to run a mile and four minutes. And IT wants the first person did IT, then everyone just start doing that. So I just to show that is possible.

That paves the way like anyone can not do IT. That's the thing that's actually possible. You don't need to do the attempt of movement.

He got the right as crazy is crazy. Crazy for people .

who don't know, can you explain how the link APP works? You have an amazing stream on the topic. Your first dream, I think on eggs are describing the APP he just described how .

works yeah so it's just an APP that nearly created um to help me interact with a computer. So on the link up, uh there are a few different settings and different um modes and things I can do on IT. So there's like the body mapping we kind of touched on.

Um there's a calibration um calibration is how I actually get cursor control. So calibrating what's going on in my brain to translate that in the critical control。 So IT will pop out models， what what they use I think is like time.

So that would be you know five minutes. And calibration will give me um so good of a model. And then five minutes for ten minutes and fifteen minutes, uh, the models will progressively get Better. And um so you know the longer I minute, generally the um Better the models would get that .

really good because you often refer to the models, the models, the thing that's constructed once you go to the calibration that and then you also talked about sometimes you'll play like a really difficult game like snake, just to see how good the model is.

Yeah yeah. The snake is kind of like my limit test for models. If I can control snake decently, well, then I know I have a pretty good model.

So yeah the link kapp has all those as web grading IT. Now um it's also how I like connect to the computer just in general. Um so they've given me a lot of like voice controls with IT at this point, so I can you know say like connect or implant disconnect. And um as long as I have uh that charger a handy that I can connect to IT. So the charger is also how I connect to the link APP.

To connect the computer, I have to have um the implant charger uh over my head when I want to connect to have to wake up because the implants in hybrid mode, like always when i'm not using IT um I think there's a setting to like wake IT up every you know so long so we could set IT to half an hour, five hours or something if I just wanted to wake up. Um so yeah i'll like connect the link APP and then go through all sorts of things, calibration for the day, maybe body mapping I have like I made them give me like a little homework tab um because I and very forgetful and I forget to do things a lot. Um so I have like a lot of data collection things. Uh that they want me to do .

is the body mapping part of the data collection .

is that also called the yeah IT is is something that they want me to um daily which i've been slacking on because i've been doing so much media traveling so much so i've been i've been super yeah i've been a terrible um first candidate for how much i've a been slacking on my homework um but yeah just something that they want me to do everyday to you know track um how uh well the neural ink is performing over time and have something to give, I imagine to give to the F D. A, you know, create all sorts of fancy charts and stuff and show like, hey, this is what the nurlan is performing, you know, day one versus day ninety versus day one eighty and things like that.

What's the color that is? That is IT like, move left. Move right.

It's a bubble game. So there will be like yellow bubbles that pop up on the screen at first. IT is open loop. So open loop, this is something that I still don't fully understand. The open loop and closed loop thing me bless .

talked for a long time about the difference beam, the two from on the technical side. So be great to hear .

europe your side of the story. Open loop is basically um I have no control over the curse um the curse will be moving on its own across the screen and I am following by intention um the curse to different bubbles and then my the algorithm is training off of what like the signals is getting are as on doing this, there are a couple different ways that we've done IT, they call IT center out targets.

So there will be a bubble in the middle and then eight bubbles around that. And cause will go from h the middle a to one side, the middle to left back, middle to up the middle, like up right. And we'll do that all the way around a circle and I will follow that cursor um the whole time and then you will train off of my intentions, what IT is expecting my intention to be um throughout the whole process.

Can you actually to when you say follow, yes, you don't mean with your eyes, you mean with your intention .

yeah so uh generally for calibration i'm doing attempted movements because I think IT works Better. I think the Better models as I progressed through calibration make IT easier um to use. Imagine moving so .

calibrated an attempted movement will create a model that makes you really effective for you to then use the force.

Yes, i've tried a doing calibration with imagined movement and IT just doesn't work as well um for some reason. So that was the center out targets is also won where you know a random target will pop up on the screen and it's the same. I just like move I follow along um with whether the curse is to that target all across the screen.

Um i've tried those with imagine movement and for some reason models just don't. They don't give its high levels quality when we get into close loop. Um I haven't played around of the the time.

So maybe like the different ways we are doing calibration now might make IT a bit Better. But what I found is there will be a point in calibration where I can use imagine movement before that point IT doesn't really work. So if I do calibration for forty five minutes, the first fifteen minutes I can use imagine movement IT just like doesn't work for some reason. Um and after a certain point I I can just sort of feel that I can tell IT moves different ah that's the best way I can I can describe IT like it's almost as if IT is anticipating what I am going to do again before I go to do IT. And so using attempted movement for fifteen minutes, at some point I can kind of tell when I like move my eyes to the next target that the curse is starting like pick up, like it's starting to understand it's learning, like what i'm going to do.

So first, it's really call that, I mean, you are true pioneer and all of this you're like expLoring how to do every aspect of this most effectively. And there's just, I imagine, so many lessons learned from this. So thank you for being a pioneer. And all these kinds of different like super technical is and it's also cool to hear that there's a different my feeling to the experience when is calibrated in different ways like just because .

imagine .

your brain is doing something different and that's why there's a different feeling to IT and then try and define the words in the measurements to those feelings would be also interesting by the end of the day. You can also measure that your actual performance, whether it's snake or web grad, you can see like what actually works well. And you're saying for the open loop calibration, the attempted movement works best for now. So the the open loop, you don't get the feedback as something that you did something .

yeah i'm strating.

No, no.

that makes sense to me like um we've done IT with the cursor and without a cursor in open loop. So sometimes it's just um say for like the center out the um you'll start calibration with a bubble lighting up and I pushed towards that bubble and then when that bubble when it's push out that bubble for save three seconds a bubble pop and then I come back to the middle um so i'm doing IT all just just by my intentions like that. What is learning anyway? So that makes sense that as long as I follow what they want me to do, you know, like follow the elaborate road that IT a all workout .

you for a great references um is that the bubble game? fun. I god.

they always feel so bad making me do calibration like, uh, we're about to do, you know, forty many calibration oh my god, which you guys want to do, two of them. My time is asking the like, whatever they need. I'm more than happy to do.

And it's not it's not bad like I get to lie there and or sit my chair and like do these things with some great people. I get to have great conversations. I can give them feedback.

Um I can talk about all sorts of things. Uh, I could throw something on on my T V in the background and kind of like split my attention between them um like it's not bad at all. I can .

Better on the I would .

love that um I love writing down suggests .

from now and that make you more fun came yeah that's one thing .

that I really, really enjoy about web ride is because i'm so competitive um like the higher the B P S, the higher the score I know the Better i'm doing.

And so if I I think i've asked at one point one of the guys like if he could give me some sort of numerical feedback for calibration, like I would like to know what they're looking at like oh, you know IT is um we see like this number while you don't calibration and that means at least on r and that we think calibration is going well um and I would love that because I would like to know if what i'm doing is gone well. Not but then i've also told me like yeah not necessarily like one to one IT doesn't actually mean that calibration is going well in some ways. Um so it's not like a hundred percent and they don't want to like what i'm experiencing or want me to change things based on that.

If that number isn't always accurate to like how the model will turn out or how like the amazon, that's at least what I got from IT. Uh one thing I do I have asked them in something that I really enjoy um striving for is towards the end of calibration. There is like a time between targets um and so I like to keep like at the end uh that numbers low as possible. So at the beginning IT can be you know four, five, six seconds between me popping bubbles, but told the end i'd like to keep IT below like one point five, or if I could get IT like one second between like bubbles because in my mind that translators um really nicely does something like web grid where I know if I can hand a target, uh one every second that i'm doing real real well.

there you go. That's a way to get a score on the calibrations like the city. Very quick.

Get from bub bubble. Yeah uh so there is the open loop and then I go to the close loop. Close loop can already start giving your sense because you're getting feedback of models.

Yes, the close loop is when I um first get current control and how they've described IT to me someone who does not understand this, I am the dumas person in the .

room every time .

I need yeah is that I am closing the loop. So I am actually now um the one that is like finishing the loop of whatever this loop is. I don't even know what the loop is.

We've never told me. They just say there is a loop and at one point is open and I can control and then I get control and is close. So i'm finishing the loop.

So how on the conversion you take? So like ten, fifteen.

well, they are trying to get that number down pretty low. That's we're even working on a lot recently is getting that down is a low as possible. So that way, you know if this is something that people need to do on a daily basis or some people need to do on a um like every other daily basis or once a week, they don't want people to be sitting in calibration for long periods of time.

I think they wanted to get IT down seven minutes or below um at least where we're at right now. I would be nice of thing you never added your calibration um so we will get there at some point. I'm sure the more we learn about the brain and like I think that you know the dream um I think right now for me to get like really, really good models. Um i'm in cela ration forty or forty five minutes um and I don't mind like I said, there was feel really bad but if it's gonna me a model that can like break these records on web d i'll stay in IT for flip in two hours.

Let's talk business so web great. Um I saw a presentation that were bless set by march. You selected eighty nine thousand targets and website. Can you explain this game? What is web grad and what is IT take to be a world class performer and web grad as you continue to break world records?

Yeah um.

Take a gold man wo yeah .

you know i'd like to think I like to think everyone who help me get here, my coaches, my parents, I drive me to practice every day five in the morning um like think god um and just overall my dedication to .

my interviews with the air, they always .

like that exact time yeah so so so .

web grade web great is great. South.

it's it's literally just agreed. They can make IT as bigger, small as you can make a git. A single box on the grid will light up and you go and click.

And IT is a way for them to match, mark, how good A B, C, I is. So it's, you know, pretty shameful. You just click targets.

only one blue cell appears, and you supposed to move the mouse to there and click on IT.

So I I like playing on like bigger grids because IT the bigger the grid, the like more B P S. It's a bits per second um that you get every time you click one. So i'll say i'll play on like thirty five by thirty five. Um great and then one of those little squares, a cell college target, whatever will light up and you move course there and you click IT and then you do that um forever.

And you have been able to achieve at first eight but first yeah and .

eight point five right now I would have beat that literally day before I came to Austin but I had like A I don't like a five second leg right at the end and I just had to wait until the len cy come down and then I kept clicking but um I was at like eight point zero one and in five seconds leg and then the next like three targets I click all stated eight pots, zero one so if I would have enable click um during that time of black I probably would have hit I don't know, I want to hit nine so i'm there i'm like i'm really close and then this whole Austin trip has really gone in the way of my web grid playing ability.

对 yeah so that you think about .

right yeah I know I just I just want I want to do Better nine I want to do Better。 I want to hit nine I think why no nine is very, very achievable and right there um I think ten I could hit maybe in the next month like I could do IT probably in the next few weeks if I really pushed I think you .

and you on a basis of the same person because last time I did the pocket with, he came in extremely frustrated that he can't beat ub illa as a droid. There was like a year ago. I think I forget like solo. Yeah and I could just tell there is some percentage of his brain entire time was thinking, like, I wish I was right now .

attempting .

I think you he did IT that night. Yeah, he stayed up and did IT that night. Yeah, just crazy to me. I mean, fundamental, really inspiring. And what you're doing is inspired in that way because, I mean, not just about the game, everything you're doing there has impact. By striving to do well on well grid, you are helping everybody figure out how to create the system all on like the decoding, the software, the hardware, the collaboration, all of IT, how to make all that work so you can do everything else really well.

Yeah I just really fun but .

that's also this part of the thing is making a fun.

Yeah it's a dick. I'm i've joked about um like what they actually did when they went put this thing in my brain. They must have flip to switch to make me a more acceptable to these kinds of games to make me a to like web create .

or something do you know .

bless this high school yeah said in seven .

point one or one .

yeah he told me.

like, doesn't on the floor with peanuts t he like fast IT. Sounds like cheating, sounds like performance enhancing.

like the first time no. And played this game, he asked, we know how could at this game and I think you me right know I want .

to get there today. I think that I think I can. I think, yeah. So i've been playing first stuff with the dweller, which really hampers my web grade playing ability. Basically, I have to wait point three seconds for every .

click so you can do the clicks. Yeah, you have to so you click by why you said .

point three point three seconds which which sucks IT really slows down how much i'm able to like how how I am able to get. I still hate like fifty. I think I have like fifty something trials net trials per minute in that um which was pretty good because unable to like there's one of the settings is also like how slow you need to be moving in order to initiate click, to start click.

So I can tell sort of when i'm on that freehold to start initiating a click just a bit early. So i'm not fully stopped over the target. When I go to click doing IT like on my way to the target a little to try to time are .

slowing down .

yeah just just a hair report.

The targets.

This is like a lead performance of okay, but that's still it's it's sucks that there's a ceiling of the point three.

Well, there I can get down the point to one point one point ones OK. Yeah, and i've played without a little bit too. I have to just a ton of different promoters in order to play with point one. And I don't have control over all that on my end yet IT IT also changes like how the models are trained, like a light train, a model, like a web grade, like a boot strap on a model s, which basically is them, uh, training models as am playing web grad um based off of like the web grad data.

And i'm so like I play web grad for ten minutes, they can train of that data specifically um in order to get me a Better model um if I do that with point three versus point one, the models come out different um the way they interact this is much, much different. So I have to be really careful. I found that doing IT with point three is actually Better in some ways unless I can do IT with point one and change all of the different promoters than that. More ideal cause obvious ly point three fashion point one so um I could I can get there. I can get there.

Can you click using your brain .

right now? It's the hover clicking with the doll curser. Um we before all the thread attractions stuff happened, we were calibrating clicks, left click, right click. That was um my previous ceiling. Um before I brought the record again with the dull curser was I think about thirty five by thirty five grade with left and right klick and you get more um P P S more bits per second using multiple clicks because it's more difficult.

Oh because what is that the book you get you're supposed to do either a left like or like right like.

So yeah blue targets for left league, orange targets for right lake is what they had done. So my previous record of seven point five with the blue and the orange target, um I think if I went back to that now um doing the click celebration, I would be able in being able to like in a click on my own, I think I would break that ten ceiling like in a couple .

days max like yeah you would start making .

bliss nervous about seventeen exactly.

Ah so what would you feel like with attractions that there is? Some of the threats .

are it's sucked IT was really, really hard the day they told me was the day of my big learning tour at the freemont facility. And they told me, like, right for one over there. IT was really hard to hear financial reaction.

Was I going fix IT? I going take IT out and fix IT. The first surgery was so easy, like, like I went to sleep couple hours later I woke up and here we are um I didn't feel any pain, didn't take like any um pain pills anything.

So I just knew that if they wanted to they could go in and put in a new one like next day if that what I took because I just wanted I wanted IT to be Better and I wanted not to lose the capability. I had so much fun um playing with IT for a few weeks, for a month I had I get IT open up so many doors for me and open up so many more possibilities that I didn't want to lose that after a month. I thought IT won't been a cruel twisted fate if I had gotten to see the view from like the top of this mountain and then have at all come crashing down after a month.

And I knew, like, see the top of the mountain. But I like I, how I saw I was, I was just now starting to climb the mountain, and I was like, there was so much more than I was possible. And so to have all of that be taken away was really, really hard.

But then on the drive over to the facility, don't know like five minute drive whatever is um I talked with my parents about IT I pray about IT I wish like you know, i'm not gonna this ruin my day i'm gonna let this i'm really this amazing a tour that they have set up for me like I want to go show everyone how much I appreciate all the work they're doing. I want to go like meet all of the people who have made this possible and I wants to go have one of the best days of my life and I did and I was amazing and IT absolutely was one of the best days i've uh ever been privileged to experience. And then for a few days, H I was pretty down on the dumps.

But uh for like the first few days afterwards, I was just like I don't know if he was going ever gonna again and then I just I made the decision that is even if I lost the ability to use the newer link, even if I lost um even if I like lost out on everything to come um if I could keep giving them data in any way, then I would do that if I need to just do um like on the daily collection every day, your body mapping every day for a year, then I would do IT um because I know that everything i'm doing helps everyone to come after me and that's all I wanted. I guess the whole reason that I did this was to help people, and I knew that anything I could do to help I would continue to do even if I never got to use the cursor again. Then, you know, I was just happy to be a part of IT.

And everything that i'd done was just a perk. IT was something that I got to experience. And I know how amazing is gonna for everyone to come after me. so. Might want to keep tracking along.

You know that's to get to work your way up to get the performance back. So as I go from rocky one or rocky, too. So when did you first realize that this is possible? And what gave you sort of the strength of motivation, determination to do IT to increase back up and be your previous record?

Um yeah was within a couple of weeks like again.

this is just like i'm interviewing on athlete. This is great. I like thing parent.

The road back was long, hard from of many difficulties there were dark days IT was a couple of weeks. I think then there is just a turning point. I think they switched how um they were measuring um the neuron Spike of my brain like the bliss helpme uh yeah the way which you are measuring and the .

behavior individual the yeah so switching from sort of individual spect detection to something called Spike power, if you watch the previous experiment to with other method J I some yeah good.

So when they did that, he was kind of like, uh, you know light over the head like like the moment, like of this works and um this seems like like we can run with this. And I saw the um up taking performance immediately.

Like I could feel IT when they switched over else like this is Better, like this is good, like everything up till this point for the last few weeks last like whatever, three or four weeks because was before they even told like everything before this sucked like, let's keep doing what we're doing now and at that point IT was not like, oh, I know i'm still only yet like saying web grade terms like four or five B P S. Compared to my seven point five before. But I know that if we keep doing this, then like, I can, I can get back there.

And they gave me the darker, and the dweller sucked at first. Not, obviously not what I want, but IT gave me a path forward to be able to continue using IT and um hopefully to continue to help out and so I just ran with IT never looked back. I guess that I just kind of person I role the punches anyway.

So what was the process? What was the feedback group on the feeling out how to do spect detection with that would actually work? Well, no.

Yeah, the good question. So maybe just describe first how the actual update work is basically update to your implant. So we just over the year, software update to his important and what you do, update your test low, your iphone and, uh, that firm will change.

Enabled us to recall averages of populations of neon nearby dual core. So we have a last resolution about which individual on is doing what, but we have a broader picture, what's going on nearby electrical and the but I mean, basically as known in scarlet was immediate when we foot that sits. Uh, I think the first day we did that, you had three for B, P, S, all of the box.

And that was A A light ball moment for, okay, this is the ipad to go down. And from there, there is A A lot of feedback around like how to make this useful for independent use. So what we care about ultimately is that you can use IT depending to do tee you want, and to get to that point, to require us to be engineer the U.

S. As you go to the dual care, to make IT something that you can use independently without us need to be involved of the time. And yeah, this is obviously story of the journey. So hopefully get back to the places where you're doing multiple clicks and using that to control much more fu everything and much more naturally, the applications .

that you try to interface with get that web grade number.

Yeah, yes.

yeah. So how's the on the harvard click IT do accidentally click self. Sometimes you like what's how hard is IT to avoid accidentally ally clicking.

I have to continuously keep IT moving basically. So like I said, there's a threshold where will initiate click. So if I ever um drop below that, it'll start and i've point three seconds to move IT before IT clicks anything um and if I don't want you to ever get there, I just keep a moving at a certain speed and like this constant, like doing circles on screen, moving a bag and forth, keep IT from clicking stuff.

Um I actually noticed A A couple weeks back that I was when I was not using the implant. I was just moving my hand back and forth or circles like I was trying to keep the curse from clicking and I was just doing IT like while I was trying to go to sleep. And also like, this is a problem.

the way clicking, I guess, does that create problems like when you're gaming accidently .

click thing like yeah yeah happens in just i've lost, i've lost a number of games because I will accidently click something .

I think the first time .

ever be us because of .

it's a nice excuse, right yeah you know any time we lose you could just say, yeah yeah you said the APP improved a lot from version one. When you first start using IT, IT was very different. So can you just talk about the trial in an air that you went through with the team like two hundred plus pages of notes? I whats that process like of .

yeah work .

going back and forth and working together to improve. The thing .

is a lot of me just using IT like day in and day out and saying, like, hey, can you guys do this for me? I give me this. I want to be able to do that. Um I need this. I think a lot of IT just doesn't occur to them maybe until someone it's actually using the APP, using the implant, it's just something that you they just never would have thought of or um it's very specific like me.

Maybe what I want is something i'm a little worried about with the next people that come is you know um maybe they will want things much different than how i've set up up or what the advice of given the team and they're gona look at some of the things i've they've added for me and like that's a dumb idea like why would he ask for that? Um and i'm really looking forward to get the next people on because I guarantee that they're going to think things that i've never thought of, they're gonna think of improvements. Not like, well, that's a really good idea like I wish I would thought of that.

Um and then they're also gonna a give me some push back but like that what you are asking them to do here um that's a bad idea. Let's do IT this way and i'm more than happy um to have that happen but is just a lot of like you know different reactions with different games applications. Um the internet is with the computer in general.

Um there's tons of bugs um that end up popping up left, right center. Um so it's just me trying to use that as much as possible and showing them what works and what doesn't work and what I would like to be Better. And um then they take that feedback and they usually create amazing things for me.

They solve these problems in ways I would have never imagined. They're so good, everything they do. Um and so i'm really thankful that I am able to give them feedback and they can make something of IT.

A lot of my feedback is like, really dumb. It's just like I want this. Please do something about IT and come back super well thought out and is way Better than anything I could have ever thought, ever implemented myself. So there's just great. They're really.

really cool as the B, C, I. Community grows. Would you like to hang out with the other folks with newer links?

Like what what relationship of would you want to have with them? Because you said like they might have a different stead of like ideas of how to use the thing. yeah. Would you be intimate by their web? Great performance.

No no I hope compete. I hope they one they like wipe the floor with me. I hope they beat IT um and they crush IT will double IT if they can um just because on one hand it's only going to push me to be Better because i'm super competitive.

I want other people to push me I think that is important for anyone trying to um achieve greatness is they need other people around them who are going to push them to be Better and I even made a joke about on x once like once the next people get join like q body caught music like i'm just excited to have other people to do this with and to like share experiences with. I'm more than happy and act them as much as they want, more than happy to give them advice. I don't know what kind of advice I could give them, but if they have questions or more than happy.

what advice would you have for the next participant and clinical .

trial that they should have fun with this a because IT is a lot of fun um and that I hope they work really, really hard because it's not just for us, it's for everyone that comes after us. And you know come to me if they need anything and to go to learning if they need anything, man nearly moves mountains like they do absolutely anything for me that they can and it's an amazing support system to have a IT puts my mind ease um for like so many things that I had, like questions about so many things I want to do. They're always there and that's really, really nice and so I just I would tell them not to be afraid to go to nearly link with any questions that they have, any concerns, uh, anything that you know they are looking to do with this and any help the nearing is capable of providing, I know they will um and I don't know I don't know, just work your off because it's it's really important that we try to give our all to this .

to have fun and work hard .

yeah yeah very good. That's not to start saying that people have fun.

work hard. Now you're real prove thly just keep IT sure maybe it's good to talk about what you've been able to do now that you have a united point like the the freedom you gain from this way of interacting with the outside world, like you play video games online and you do that by yourself yeah and that's a kind of freedom. Can you speak to that freedom again?

He had what all I know, people in my position, what they just want, more independence. The more low that I can take away from people around me, the Better if i'm able to interact with the world without using my family, without going through any of my friends um like needing them to help me with things. The Better um if i'm able to set up on my computer all night.

And not need someone to like sit me up a say like on my ipad, like in a position where I can use IT and then have to have them wait up for me all night until i'm ready to be done using IT. Um like that IT takes a load off of all of us and it's it's really like all I can ask for um something that you know I could never think nearly enough for. I know my family feels the same way. You know, just being able to have the freedom to do things on my own at any hour of the day, night IT means the world to me. And I don't know when you're .

up to A M playing web good by yourself yeah I just imagine like it's darkness and just a light glowing and you're just focus what what's going to your mind. Are you like in a state of flow where it's like the mind is empty.

like those like them masters yeah generally IT is me playing music of some sort of a massive playlist. And so i'm just like rocking out to music and then it's also just like a race against time, because I can constantly looking at how much battery percentage I ve left on my implant.

Like art, I have thirty percent, which equates to you know x of time, which means I have to break this record in the next you know hour and half rose is not happening tonight um and so it's it's a little stressful and when that happens, when it's like when it's above fifty percent, i'm like, okay like I got time. It's not going down to thirty and then twenty is like alright at ten percent a little pop ups s gona pop up. But here it's gona really screw my web grade flow.

It's gonna me that you know like there's a like a low battery, low battery pop up comes up and it's really going to scream over sell if I have to, if I going to break this record, I have to do IT in the next like thirty seconds also that pop up gonna get in the way I cover my web grad um and then after that I go click ic go back in the web grad and like I that means I have you know ten minutes left before this thing is dead that's what's going on to my head generally that in whatever songs playing um and I just I just want I want to break those records so bad like it's all I want when i'm playing web group IT has become less of like now this is just a legally activity like I just enjoy doing this because I just feels so nice and that puts me at IT is no once i'm in web could you Better break this record or you're going to waste like five hours of your life right now? And um I don't know. It's just fine. It's fine. Man.

have you ever tried web grade with like two targets and three targets? Can you get higher B, P, S.

With that? You can you do that .

even like different color .

targets or even possible target?

yes. So B, P, S is a log of number of targets times, correct, incorrect, divide time. And so you can think different clicks basically double the number of active targets.

good. So know you basically hired B, P, S to more options. Are the more difficult to task. And there's also example you ve played in before, which is, yes.

I think the t is with a and why yeah and .

so you can go like that saying, yeah.

he isn't like a because identity B P S.

So like you know, I then .

put in a giant B, P, S in the background. So now it's like the opposite of the mode. It's like like super hard mode, metal mode of this, just like .

a so you .

also play civilians. Six.

I love.

usually go with korea. I do.

So the great part about korea is they focus on like science tech Victories which was not plan like i've been playing career for years and then all of the nearing stuff happened um so I kind of aligned um but what i've noticed with tech Victories is if you can just rush to rush science um then you can do anything like at one point in the game, you will be so far ahead of everyone technologically that you will have like basket men, instance men, plane sometimes and people will still be fighting with like bozen errors. And so if you want to win a domination Victory, you just get to a certain point with the science and then go wipe out the rest of the world or um you can just take science all the way when that way and you're going to be so far ahead, everyone, because you're producing so much science that it's not even close. Um i've accidently won in different ways just by focusing on .

science one by fox.

I yeah I like I I was playing only science obviously like just science all the way, just tag I was trying to get like every tech in the tech tory and stuff and then I accidentally one, uh, through diplomatic Victory and I was so mad, I was so mad, uh, because I just like, ends the game one turn. I was like, oh, you won. You're so diplomatic like I don't want to do this.

I should declared war on more people or something that was terrible. But you don't need like giant civilizations with tech, special with korea. You can keep IT pretty small.

So I generally just, you know, get to a certain military unit and put them all around my border to keep everyone out. And then I will just build up. So a very isolationist .

yeah just you making that sounds so fun so much. And I also saw .

a citizen trade.

and that's probably .

seven up alpha bea test, whatever.

我 我。 What other stuff would you like to see improved about the new link cap in just the entire experience?

I would like to I guess i'd get back to the um like click on demand, like the regular clicks. That would be great. Uh, I would like to be able to connect to more devices, right knowledge, just the computer.

I'd like to be able to use IT on my phone or using on different councils, different um platforms. Um i'd like to be able to control as much as as possible, honestly, like an optimistic robot would be pretty cool. That would be sick if I could control an optimistic robot.

The link up itself um IT seems like we are getting pretty um died in to what um IT might look like down the road. Seems like we've gotten through a lot of what I want from IT at least the only other thing I would say is like more control over all the promoters that I am can tweak with my like curse and stuff. There's a lot of things that you know go into how the cursor moves in certain ways. Um and I have I don't like three or four of those premature and there .

my gain and friction .

gain friction yeah and there's maybe double even of those with just like velocity and then with the actual dual curser. Um so I would like all of IT. I want as much control over my environment as possible.

You want like advanced mode you know like like this man is usually this basic mode and you're like one of those folks like the that's .

that's that's what I want. I want as much control over this is possible um so yeah that's that's really all I can ask for. Just give me give me everything uh.

as speech been useful as he was being able to talk also in addition everything else yeah you .

mean like while i'm using IT.

while you're using IT like speed to texts, oh yeah, what do you type or the also keyboard yes yeah .

it's a virtual keyboard. That's another thing. I would like to work more on finding some way to um type or text in a different way right now.

IT is um like a dictation basically and a virtual keyboard that I can use with the cursor. But we've played around with um like finger spelling, like some language finger spelling um and that seems really promising. So I have this thought in my head that is going to be a very similar learning curve. But I had with um the curse where I went from attempted movement to imagine movement at one point, I have a feeling um this is just my intuition that at some point i'm going to be doing finger spelling and I won't need to actually attempt to finger spelling anymore, that i'll just be able to think the like letter that I want and IT .

that will be happy and that's chAllenging. It's hard. There's a lot work for you to take that leap without be awesome. And then like .

going from letters to words is another step like you would go from you know right now it's finger spelling, like just the sign language hala bit. But if it's able to pick that up, then I should be able to pick up like the whole sign language like language um and so then if I could do something along these lines or just the sign language um spelled word, if I can you know spelling at a reasonable speed and I can pick that up, then I would just be able to think that through and it'll do the same thing. I don't see why not after what I saw with the um with the current control, I don't see why IT wouldn't work, but we'd have to play round of them more.

What was the process in terms of like training yourself to go from attempted movement to imagine movie, how long did they take? So how long will this kind of process take?

Well, IT was a couple of weeks before IT, just like happened upon me. But now that I know that that was possible, I think I could make that happen with other things, I think would be much, much simply.

Would you get an upgraded implant to this?

sure. absolutely. When you whenever theyll .

let me so you don't have any new concerns for you to surge experiences all IT was I no regrets?

No.

sorry.

Everything has been good so far.

You so you just keep you a nub grades.

yeah. I mean, why not? I've seen how much its impact on my life already and I know that everything from here on a ledge gona get Better and Better. So um I would love to I would love to get the upgrade.

What future capabilities are you excited about sort of beyond this kind of telepathy? Is vision interesting for a folks, for example, how blind so you are like a enabling people to see our offer speech?

Yeah, there's a lot that's very, very cool about this. Mean, we're talking about the brain. So there's like this is just motor cortex stuff. There's so much more that can be done. The vision one is fascinating to me.

I think that is going be very, very cool to give some of the ability to see for the first time in their life would just be unit IT might be more amazing than even helping someone like me like that just sounds incredible. Um the speech thing is really interesting. Being able to have some sort like real time translation and um cut away that language barrier would be really cool.

Um any sort of like actual impairments um that he could solve like with speech would be very, very cool. And then also there are a lot of different disabilities that all originate in the brain and you would be able to hopefully be able to solve a lot of those. Um I know there's already stuff to help people with seizure um that can be implied in the brain.

This would do imagine the same thing and so you could do something like that. I know that you know even someone like joe rogan na talked about a the possibilities with being able to i'm stimulate the brain in different ways. Um i'm not sure i'm not sure what you know like how ethical a lot of that would be.

That's beyond me honestly. But I know that there is a lot that can be done when we're talking about the brain and being able to go in and physically make changes to help people or to improve their lives. So i'm really looking forward to everything that comes from this and I don't think it's all that far off. Um I think a lot of this can be implemented within my lifetime um assuming that I live a long life.

what you're referred to, things like people suffering from depression or things of that nature potentially getting help .

yeah flip switch like that makes someone happy. Um I know I think joe is talking about IT more in terms of like 1万the experience like what a drug trip feels like like you want to experience what you like to beyond course。 Yeah, mushrooms sound like that.

D M, T, like you can just fill that switch in the brain. And my body being is talked about being able to like a White parts of your memory. And we experience things that like for the first time, like your favorite movie or your favorite book, like just write that out or quick, and then we fall love with Harry potter or something. I told him I was like, I don't know how I feel about like people being able to White part of your memory. That seems a little sketch to me, is like the ready .

doing IT so sounds that would love memory replay just like actually like .

high resolution replay memory.

He has a black mare who always can consider the worst case, which is important. I think people don't consider the best case of the average case enough. I don't know what is about us humans.

We want to think about the worst possible thing. We love drama. Like how's this new technology going to kill everybody? We just love that. Yes.

watch hopefull. People don't think about that too much. Me, little room in a lot of my plans .

assume you're going out to take over the world. I mean, I love your twitter. You you tweeted. I'd like to make jokes about hearing voices in my heads since getting the neural link, but I feel like people would take a through away, plus the voices in my head told me not to, yeah, yeah, yeah. Please never stop. So you were talking about optimist, is that something um you would love to be able to do to control the A M of the entire of the .

you think there's something like fundamentally .

different about just being able to physically interact .

world yeah one hundred and um um this I I know another thing with like being able to like give people the ability like feel sensation in stuff to by going in with the brain and having a neri c maybe do that, that could be something that um could be translated through transfer through the optimists as well like there's all sorts of really cool i'm interplay between that and then also like such as physically interacting.

I mean ninety nine percent of the things that I can't do myself um obviously need I need a caretaker for someone to physically do things for me if an optimistic robot could do that like I could live an incredibly independent life, not be such a burden on those around me um and that would they would change the way people like me live um at least until whatever this is gets cured um but being able to interact with the world physically like that would just be amazing um and and they're not just like for being for having A B A C taker. Something something like I talked about just being able to read a book imagine opting is about just being able to hold a book open in front of me. I get that smell again.

I might not be able to feel IT at that point um or maybe I could again with the sensation and stuff but being there's something different about reading like a physical book then staring at a screen or listened to an audio book actually don't like audio books. I've listened to a ton of them at this point, but I don't really like them. I would much rather like read a physical coffee.

So one of the things you would love to be able to experience this opening the book, bringing IT up to you yeah, and to feel the touch of the .

paper yeah, how man to touch the smell and which just something about the words on the page they've replicated, you know, that page color, unlike the kindle and stuff yeah, it's just not the same. yeah. So just something as simple as that.

So one of the things you miss this touch.

I do yeah, a lot of, a lot of things that I interact within the world, like clothes, or really any physical thing that I interact within the world. A lot of times what people around we will do, they're just come like a rabbit on my face. They're like lay something on me so I can feel the weight. They will rub you know a shirt on me so can feel fabric like there's something very profound about touch and uh is is is is something that I miss a lot um and something I would love to do again will see what be the first thing you do with the .

with the hand that can touch mama hug after that right yeah yeah I know that is .

one thing that i've that i've asked like god for basically every day since um my accident which is being able to like one day move even if he was only like my hand so that way like I could squeak my mom's hand or something just like shower that you know like how much I care and how much I love in everything um something along those lines are being able to just interact with the people around me hand they give someone a hug um I don't know anything like that being able to help me eat like i'd probably get really fat um what would be a terrible。

Terrible thing uh also be blessing chest on physical test board .

yeah yeah I mean there are just so many upsides. Know any any way to find some way to feel like i'm bringing bss down to my level yeah because yeah he's just such an amazing guy and everything about him was just so above and beyond um there anything I can do to taken down or not yeah yeah .

humble in a bit. He needs IT yeah okay, as you sitting next to me, um did you ever make sense of why god puts good people through the charge?

Oh, man. I think it's all about. Understanding how much we need god and I don't think that there's any light without the dark. I think that if all of us were happy all the time um there would be, you know no reason to turn to god ever. I feel like there be no concept of, you know, good or bad. And I think that as much like the darkness and the evil is in the world, IT makes us all appreciate the good and the things we have so much more. And I think, you know, like when I had my accident, the first one of the first things I said, the one of my best friends, was, this was nothing like the first month. Two, after my accident, I said, you know, everything about this accident, I just made me understand and believe that, my god, israel, and that really is a good basically that um like my interactions within involved in you know real and worthwhile and he said if anything seeing me go to the accident, he believes that there is a god and it's a very different reaction um but I believe that IT is IT is a way forgot to test us to build our character to um um and three trials and tribulations to make sure that we understand um how precious you know he is and the things he's given us in the time that he's given us and then um hopefully grow from all of that um I think that's a huge part of being here is to um not just you know have an easy life and do everything .

that .

easy but to step out of our country zones and really chAllenge e ourselves because I think that's how we grow what gives you hope about this .

whole thing? We have gone on .

human S I think people are my biggest uh inspiration, even just being a nursing um for a few months looking people the eyes and hearing their motivations for why they're doing this. It's it's so inspiring and I know that they could be other places and curse your jobs um working somewhere else doing X Y S Y that doesn't really mean that much.

Um but inside they're here and they want to Better humanity and they want to Better just the people around them. The people have interacted with their life. They want to make Better lives for their own family members who might have disabilities.

Or they look at someone like me and they say, you know, I can do something about that. I'm going to. And so has been what i've connected with most in the world.

people. I i've always been a people person, and I love learning about people. And I love learning like how people developed and where they came from and to see like how much people are willing to do for someone like me.

When they don't have to, they're going out of their way to make my life Better. IT gives me a lot of hope for just humanity in general, how much how much we care and how much we're capable of when we all kind of get together and try to make a difference. And I know there's a lot of bad out there in the world, but there always husband, there always will be.

Um and I think that that is IT shows human resiliency and IT shows what we're what we're able to endure and how much. How much we just want to be there and help each other and how much satisfaction we get from that because I think that's one of the reasons I were here is just to help each other. And um I don't know that that always gives me hope is just realizing that there are people out there who still care and who want to help .

and thank you for being one such human being and continue to be a great human being through everything has been through and being an inspiration to many people to myself for many reasons including your epic unbelievably great performance on web. God, I will be training all night tonight to to .

try catch you can do IT and I believe .

in you that you can, uh, once you come back so sorry to interact with the Austin trip once you come back when SHE beat plus yeah yeah solute.

i'm looking for you. The whole .

world is rooting for you. Thank you for everything you ve done.

Man, thanks, thanks.

Thanks for listening to this conversation. We know an arba. And before that, with ella musk, D.

J. Saw Matthew google and base chapman to support this pogue's. Please check out our sponsors in the description.

I now let me leave you some words from oldest hux ly. In the doors of perception, we live together. We act on and react to one another.

But always, and in all circumstances, we are by ourselves, the matters go hand in hand into the arena. They are crucified, a loan embrace. The lovers desperately try to fuse their insulated excise into single self, transcendent in vain.

But it's very nature. Every inboard y's spirit is doomed to suffer and enjoy its solitude, sensations, feelings inside fancies. All these are private.

And except through symbols and a second hand incommunicable, we can pull information about experiences, but never the experiences themselves. From family to nation, every human group is a society of island universes. Thank you for listening and hope to see you next time.