Ep 100: Why AI Is Underhyped with Elad Gil
50:23 Share

R L, M, S going to expansion, ally, keep getting Better to the point that they like total .

changes improves IT? Or are they hitting some kind of the way I think about IT is almost like L M steps, right? And when you had GPT three point five and you went to four, suddenly legal open up is a vertical. I back the company called Hardy, and they showed me side.

side, two, five.

This doesn't work, right? You had a model to take something you could do legal. What's the next step? pocatello? I think about what you could do with like GPT one, you know you have these steps. And so is this latter that you're climbing and adventure. You're going to hit a point where can do all of human services.

We have something special for episode one hundred of american optimists. Ela girl has been a friend for a long time, is one of the great thinker as great investors and silicon valley, we go really deep on A I on history, of the tech world, on what's coming up next in the next ten, twenty years. Or e lads funds.

As one of the funds I am the most bullsh on. He just involved in so many the best companies over the last twenty years, and already probably involved in things that will become the most famous companies in next decade. Excited for to meet.

I'm joe long zyl ll come to american optimists. Have my friend go with us and really glad you here. Things coming.

Oh, thanks. Great to see. So what's all your background? Tell a little bit about your brain .

how you make IT. Sorry, I moved out to the bay area. Reason for a job. I going to start up um sort of right at the end of the telecom are the dot com boom and sort of bus and enforce.

I had perfectly bad timing and I showed up as everything's collapsing and I worked at a telecom equipment start up that got um that went through a series of lady s kind of laid off in the bubble. And we grew from one hundred and one hundred people, three to thirty people, still one. Now I got cut around fifty years, so thousand in the third round. So I made the through two rounds and but is a pretty vital enviroment. Everything was kind of collapsing. It's actually very good as very formative, right? Because if you've been through a layoff and a really tough time like that and I straight at the schools money and so I remember um even as that company was ramming up, I was like this is unsustainable like you I don't know how this thing is onna survive and so I remember me in a friend and I used to go to the grocery restore um every week once a week and we buy like a low of bread and I think that cheese and that's I just see chee sandwiches every meal like OK because I was like saving money because I thought the slave was coming in the coming in the library you know something yeah look at their burn and you look at their cause you to do how do you .

know most engineers aren't like aware those financial things do you know to be ah and I just felt .

like math to me honestly like it's just you start thinking about IT, like these things seem off and then you do the math. I know like were going to other money and five months or whatever, they don't do something, things like to do something. And you know if it's it's kind of funny how if you just look at certain things from first principles or you just look at data you're like of, of course.

this thing is this way. Did you become an if you get laid off now.

I ended up my way to google. google. The time was sort of scaling, and IT was soaking in all the talent in silicon body, at least the entrepreneurs ent.

And so I joined there and um I I ended up working on two things, one as I held by in the android team and start a lot of the only mobile efforts. And then second, I worked on some early um AI machine learning for ads targeting related products. So I worked more on the product side um and then after that I would have to start a company and um sold that to twitter. And then I worked to twitter for bunch years and then start in other companies.

So where the lessons from building a team of google on the mobile sign would you guys that I was always pretty important team for google.

You do yeah know at the time um uh Larry page um was a very activable and he decided to get rid of all middle managers and so when I joined to removed like three management layers and suddenly every director b p 5 person had fifty two hundred people working directly for them and so they had no ability to actually meet with the the team that they work for them， right? Because how can you meet with one hundred people?

Would you together with you?

Now, basically you have this giant Green mark for talent. We could do IT everyone, because nobody is paying attention, right? So is actually, for me, is great.

And so I recruit a lot of the early mobile team without their managers knowing, because their managers, you know what they were doing. So I just come and start working with me in mobile. And so we kind of organically built this. I don't want to call IT like this team of pirates or something.

Just kind of seen is this because the google is making so much I like to think of google is like this, like of money, and people like all around IT and this kind of a mess do stuff.

I mean, IT was basically a golden era where silk valley had just gone through a major depression, hugely offit happened. A bunch of companies had blown up. Founders were a service out industry, and google is soaked in all the very, very best talent of that.

But then also just let IT do whatever I .

wanted up to a point you and eventually they put back in the dle management and you know they did a budor y orbs and realigned everything but IT was a very um if you, anna, go try to start something. If you can convince one of a handful of people that you could go do this.

So you are really an tremendous .

r inside of google, basically yeah, you know, I never think that you see how hard that is to really be an entrepreneur. Well, no big company really has entrepreneurs. But yeah, was basically you had this grey market for talent.

People could just start all the things and that's when you had this big expansion in gi used to have this web page up that was like, here's all the things we're never going to do. And I said we're never gonna I am or chat. We're never going to do a browser. We're never going to doing all those things. We're never going to satellite supposed .

to be like this .

exaggerated funny .

things funny.

So yeah, so IT was a very different namic time. And I feel like at any given moment in time is one or two companies that are going to this kind of golden age period where you're expanding rapidly, you're sinking in all the best talent, you're launching multiple product and you have the diversity of stuff happening. And then often these cowards of people go off and then do amazing things.

And so a meli later. And so maybe right now that's open the eye. Maybe that's stripe as a handful of companies that are probably .

these rap in a little IT like that in some ways. But going back again, so why do you leave to start a company .

after google? Yeah, I was want to start a company is because I .

had worked at to start up that failed.

I had no money. I the the era is actually really hard to start companies, right? There weren't that many people funding IT.

There wasn't White combinator these accelerated programs. You could just go and people give me some money to get going. They weren't very many info investors that wasn't information .

online about how to do that to a sandwich ously.

right? Like I was literally paying off school that yeah and so you're very constrained and I feel very much less constraining environment now because seed capital much easier, which means that you know really smart people going to show up, get some money and start doing things that are very ug age.

In some ways though, if to be an out for back then, there was a higher bar was in there for most for lot of people.

is IT was IT definitely seems like IT was harder. And me being person now, and it's always harder.

I was Young her way up.

But that way IT was really hard to get something up in winning mac and even convince people that quit and go do IT because they weren't that many start of successes until the no um especially after the dot com bubble where everything collapses， people to go with the lesson of oh, all these things are kind of think companies .

and I want to go somewhere no, I remember hiring for arvin people what the company yeah the first company .

Harry was called mixture labs. IT basically ended up being a data in for a platform company before. There was a lot of those for mainly for developers to use or so for engineers to basically build different applications. We ended up selling IT to twitter when .

twitter was about ninety people. Twitter really need some help back.

And is my recollection twitter was a mess, right? And it's kind of funny because as you see, multiple companies go from, you know fifty, one hundred people, that thousands of people, you see that almost all of them have these phases where where a mass or something really bad is happening or you know, serial things problems had to be done with. Twitter was especially bad.

Yeah, twitter seems uniquely bad for me. The thing i've heard you could tell me maybe it's too mean, but it's like, I mean, I think mark zk work, by the way, call that a clown car that that fell to a gold mine, which a that's pretty mean, but pretty funy. The thing I heard is that i'd like to say that you don't get billion or companies without really top tech cultures except for twitters, like the exception .

that rules a rule I can think of a couple of others. So I think my big lesson over the last couple years is that product market fit matters above everything else. In other words, if there is a active market that really wants to use what you're doing IT, it'll put up with a lot of stuff.

And this there is a very AR alternative. Fawell was up constant and the think growing, right? And there would be articles about the film, and I would just a boost growth.

And so twitter was in this magic moment. What was this doing is very unique, defensive, honestly. I mean, survived to this day now x under mask, but survived for twenty years, something that wasn't very well for much of that history.

What when you came in, you guys to fix some things there.

at least we fix so much stuff. So twitter was in a state where um you meet in the feel up going and they put up a picture of very cute whale like a graphic and that was their product because you can log in or think because the site was down, because you can deal the traffic at the same time when we were bought. Um they couldn't um what's known as deploy code.

They couldn't update the website and they had been updating IT for a week. I just didn't know how to update the code. And so we went and we fixed that to, boy you, that was the first. My team was .

supose to go .

and build other stuff, the equal and in A X.

Well, how long you stay there?

I was there two thousand years full time. In the other years, an adviser, and my job was basically, if I became one of the fixture in the company, we got gotten two dozen nine, and I started off running like search and G O and some other kind of A I centric teams and then I morphed into this fixer role. Um we basically had a big um reorganize and um basically what happened as we did kind of a slow motion rework.

We want to do so look like a month to reword the product and dick a as each person what they want to be doing. And most people said, I I want this job, but I want that job and I showed and I said, i'd just tell me what you need me to you, right? Like it's not about me.

It's about helping this company. And so half the team got fired. I got promoted for saying that slash acting that way. And then I got kind of pulled into this fixer role.

And I ended up getting involved with ma.

with name and something maybe help higher in the exact.

and then go on to do the next thing. And I want to ask just because let's give me a law, people's minds. And I want to focus too much on this cultural stuff.

But like when elon bought twitter and became x, my friends went in there and there were accounts like mine that were like turn down. And I was all this crazy politics that very clear, it's very poltice ze. And I felt like that might have started a more twenty, twenty fifty. Was that going on a little bit there?

When I was there was actually viewed more as a platform for helping disseminate information for freedom.

was like free speech was more on the same .

yeah because I was that urb spring was happening on top of twitter. Um IT was starting to be uses like a news media where I don't remember there is .

that famous landing .

of the plane in the in the river yeah yeah there was a .

cool first photo on witter.

And so um in that era, there is also a lot of x google people at twitter, particularly in the legal team, and they came from this mindset, uh, what's fair use and how do you create more access to information is .

very much the all kind of like class school liberal silicon value freedom and free speech and everything. So that fast things you never experience any of other mass is all what you ve you to build.

Yeah I want to start a company um so I start a company called uh, color with you. You are thankfully help back. I'm certain appreciate all your support earlier on that. And um is basically a really, really cancer genomics and digital health company that was focused on helping people get really key information about their health, particularly around genetic risk .

of cancer and related areas. You actually a background both in math and biology. Well, obviously around twelve, there are a new possibilities coming up with genomic. What I wasn't ire you to do IT.

Yeah there is one of three points of inspiration. I think the biggest one was just my cofounder and who is now still running the company. Um he had a familiar history of breast cancer.

His mother had breast cancer twice. He had multiple family members die of IT. And so the real driving for stress was was just how do we help people get really important information key to their health? Yeah and so that the reason we started the company, we want to help people.

and you probably know this story is IT was something really inspiring for us as my my now wife we were dating at the time and and I think I so I like bow a couple lectured test. So you guys given us like like low cost and he took one and always and SHE had this like crazy risk. And twenty three and me, I told her he had no risk. And now, you know, he has his crazy cancer risk.

So just in case a her family took the test and her mother was sixty one of the time, and I came back and they said eighty percent chance of cancer by time your sixty based on based on results and this daughter said, I don't worry about that but that's really strange you would know that I never seen that before just in case probably at the age we should do may take out over ah and so three or four months later, no rush. They took him out and they found stage three cancer. yeah. And so fortunately for the real that treated at help her SHE brother wouf, when we lived a couple years, because they found and SHE lived like six or seven more years, he got to meet a couple of fever grandkids which which is really wonderful that at least he gave five years of life, thinks to one hundred dollars as it's shocking to me health care like that one hundred dollars think you just like .

prolonged someone's life a lot yeah it's amazing um how little actual preventative care helps. And by the way, the reason like what you just said is the reason we start the company. Thank you.

right? So you know, it's it's one of things that I never gonna work at anything about color, simply because about to help people in these ways what didn't go well, lessons learned from I mean, it's a health care is really hard as a industry and um there's a regulatory part of IT. Honestly, people just don't want to pay at a pocket for their health.

So do you a point, you can pay one hundred hundred dollars and get information that maybe live saving people want do that right? There used to their employer paying for a of the government paying for IT. And most of hell care is chronic versus preventive. Another, you wait for somebody to get really sick and then you try to treat them versus say, how do we just prevent all these things from happening?

To begin with, hundred. One of the one of my neighbors and who we've had on here and back in texas is Peter a tea, who's the healthcare guy and they give me these frameworks where it's like the average doctor will say your clustering is like, you know sixty th percent tile like bad, but like it's not not even something. No need to do anything.

Where is a really good doctors? Like let's get IT to be the very best, let's not just leave IT. Like okay to mediocrities. Ep pushing your health, which is is a Better framework.

There's a very basic stuff you can be doing and people just don't know about or don't know how to do. Um I mean, the other part of IT is when you look at A A lot of our focus from a biographical acute al perspective, like the new drugs wee developing things like that often IT aren't going to move in the needle that much. Actually, they're very important for specific disease.

But if you got rid of all of cancer in all of hard disease, thank you. Add something like five to seven years to the ever is a really five to is very small amount on the and like, o what's everything else? And I, should we actually be thinking about extending lifespan and health span? How do you keep people healthy for longer? And we're not really doing much to developed drugs in that direction.

And I I think there's big gaps in the market. Other things you're doing today, actually.

there's one company that I backed up called BIOS is just one public, which i'm very side. I have involved with them sense of very really days. I think they may still be on their quiet period.

I don't know there's much I can say about IT, I think at a very high level of working on a different types of drugs to help with aspects of ageing. And I think that's a very exciting area overall. And they're announced certain things as you, I think, checking them out .

totally warm and time and see my friend rebuilt bunch, these companies in the latest stuff they're doing with epps and aging. This really is like size. I really hopeful for helps.

I want to ask in general, uh, about your successes and investors. So you you obviously talk a little bit about this company as you build and you did really well them, but I think you're probably most famous these days. As for someone who's invested, I think may be dozens of you back some really top companies from the beginning.

You mentioned the strike earlier in one of the one of the famous ones. You kind of invested and really, really want to be a mattered to those guys. How did you find guys like that? How are you able to keep winning again and again and again with with these early stage investments?

Yeah I think i've been super lucky that stuff um you know I think a lot of the companies that are involved with quite early, uh, was just because I was helping our people on the equis, some like because I was starting a company myself and people to start coming to me. They're for advice or I was a couple months ahead of them or we would help each other.

You just happen to help .

the really talented ones. Yeah just fell into IT. You know I just fell into the .

a and .

the car in the goldmines, whatever. Uh, a lot of these early teams and they just happen to be some of the bigger companies, you know L, B, N, B, it's strike. You know, eventually they are helping our coin base and vi, gma and card and been a great right.

Like what is IT about teams, for example, like stripe, to differentiate them from others? Like what what about what guys running the company? Me special.

Yeah, I have been kind of noodle in on this. I had a conversation with some fix on my team, uh, yesterday. Where is this trying to bring storm? What are other specific signs of truly outside successful outcomes in terms of the types of founders? And I think a lot of IT is just like you end up in the right product market, is the market big and off, is a growing, is a dynamic in the right way. And so I think half of IT is just you find the right market and some .

of that could be a lot and some of that could. But those guys are like, I mean, obviously knew you .

are special, but they seem like they're like some of the leading kind of tons of people into so what's different? And so I almost view IT as there being three types of founders that become hyper successful. And I just yeah testing this out spit bowling IT right on this.

correct? The first one is kind of a polymathers's per intellectual, yet very competitive person. And that's Polly, Patrick, king john from strike. Honestly, that was Larry and sergey when I were to school. They were very polymax c very deep on everything theyd go down into the weeds on the data. And so I think one almost like super founder arch type is that I think the second one is the super hard core, extremely focused. Really, really driven over drive founder that may be you know, travel from uber.

There might be Peter, I wish, back in the day, but we really doing .

well at Peter was just super focus, I think, signal because a lot of people are ecosystem example, are doing Angel investing or they involve a lots of other companies while they are running their company. And that second classic founders didn't do any, that they are actually say no that everything they're .

just all on one thing.

And so I think that's another arch type. And because a few people have massive that is incredibly focused.

I like evan used to be more that way because he really would say no to everything .

yeah some of the best founders are just um like Parker replying as incredibly judicious with this time and he just like this is you know he's really focused. And then I think there's a third type of founder, which is just like the network effect business is just do you up something that has network effects? We're going to do well.

well even if even if your long can you be well.

you could be sucked and then to the next level, right? And you start doing things like long and .

things like that. So maybe has an effect .

plus some of the um maybe he's sort of an overlapping arch type of like one of those things plus you now have a network effect business. A faut fences have .

seen a lot of more works of business. We always talked about a obviously those little giant companies in our world and there there's some enterprise. But what are you seeing in that reference these days and are exciting.

Yeah i'm saying last, network affects and more scale effects, which are related or overlapping where as you get more scale, you end up with cost advantages and then allow you to get more scale. You could argue maybe stripe IT. Isn't that basketball? There's also more modern version of that, particularly in energy and other related markets.

Um so you know it's back to a broader question of how do you create defensively in a business and is almost like five or six common patterns of defensively scale effects is one network affects as another. I'm creating an ecosystem on top of your platform, like what sales force is done as one. Um there may be long term contracts and durability, those contracts which happens, for example, in the medical dirigo tion world. So is in the in the world of business business like five ways or six ways to create defensive ability and roughly, everything pulls into those.

Yeah apple, sure. We always try to create network affects, and we did get a few of them going. So we got a lot of countries working over IT with dozens of countries we can have to be on IT.

I think we are really proud to sell to israel and that was there was a big deal that they box and never buy anything for anyone. But I think that was probably that, that would expect. I like think what is good enough, but I think that kind helps. And then we had like airbus using with all the suppliers and want how to be on IT. So I think I think there are enterprising where effects you can do, but if but there they do seem to be more rare these days.

Yeah, I think so. And then there is also just the scale effects we mentioned. Like I know we were both involved with another early, early and that's a good example where they now has such a profile products. They have the customer relationships and that allows a fly away terms, the ability to continue to cross all .

different types of products. I think you get big enough to you get a reputation in dc. You can like do things in dc, you crazy on android one.

So we have this company surono. I think you're all so just build and it's grown very fast for the navy and and people they access to weapon ze, some of these Thomas vessels. And we reached out to future from primes and andero.

So good like within three days, three days the the models data exact how IT works in back. And then IT was like I took a multiple weeks even to get the nd a sign with, like the other guys. And so it's kind of cool, because the underworld is now this thing that you can use as a prime, making a grave and fashion.

because everyone wants to work with this. This is the best. Yeah, yeah. You know that, that is one of those. The feedback loop is very positive for some of these businesses.

And so I think everybody always underestimates the degree to with these feedback loops and up mattingly in the way that IT changes the velocity infrastructure of a business. And if you are up on a positive track, ket keeps going. And that's also a true for hiring. There's certain networks that you tap into the gap the five will going on. Somebody and my team actually went back.

And what are all there are common schools that all the founders of the biggest companies went to or the last decade for examples, right? And there's basically three schools that are like a huge chunk of value created and then there maybe five more, and that's very disperse. It's kind of we'd expect at this treat neural.

It's stanford. Um it's M I T. It's harvard and then there's three of the next set which is like duke it's konnie email and you know it's a handful .

of places to .

can make IT on there particularly for .

A I and watching stanford my a school curie I .

is really interesting burke. I has a lot of P H D N A I who have started .

interesting company. I actually .

have higher, quite a few P H S A A S. And so that these weird clusters or pockets of talent, they go on to do interesting things. And that's actually been true throughout history, right? You go back, yeah and you look at the .

ison so you look at paris, uh, in the late eight and hungry maf .

tisean is crazy exactly. And that's that's true for art movements. That's true for movements and technology movements. That's amazing.

A A lot of this successful i've talked about this right now is is still in your ever really popular hygrometer handbook, right? So as you published this with that came out in last year to .

come a couple years ago.

what are a few lessons of frameworks from .

that you tell us about me? I mean, because basically a book that says, say that your company starts working, how do you scale? I saw this near charms article the day we're calling cop in this, said the book sitters by his bedside right now. And my book was the first one listed out like what is amazing that it's it's kind of spreading more, brother, beyond just the technical stem that's cool. Are some of the .

high growth like strategy relevant for other areas?

Maybe yeah all about just scaling organizations dealing with um uh change in your organization. And then there has a lot of very taco stuff like how to raise money or how to deal with the bad board member or you know different aspects of really building out an organization. And the organization could be a uh, technology company IT could be an unprofitable be a variety things.

It's funny when the book came out to actually had people from all over the world reach out around different types of organizations, they were scaling ahead. And nonprofit indonesia reach out to me. I call, you know, different technology company is. So it's been a really fun, right?

That's really cool. So I get a lot of what we're doing that s apply to .

a lot other fields that I common patterns. And then I think the only good dinner started advice. There is no good genre started advice. And so I think a lot of things for any organza a situation and who's involved in, who are the people and how do you want to structure .

and everything else makes sense. Nothing I want to ask you about is I think recently interview linna on yeah this is not my favorite government leader um I think you are very applied to her. I think A A little bit about like regressive of any enforcement in my experience has been things in biotech that are being really like hurt and set back and these things not being cured because she's being so gressier blocking things um like what you think of answer about the about when you talked about I think you talked about like blocking meta from acquiring a thirty person company seemed to a little crazy to me. You should push your harder like where are you learn this?

Yeah so we have a reason. Be short interview and he was generous with our time to do this. right? Have about thirty minutes. And um I think there's a lot of standards questions that we asked an M A A I and open source and all these topics, and we just didn't have a lot of time to go deep on a lot of things.

And I want to see the things I was most interested in because actually feel that her stands on small m na by big tech is very well understood and known, right? It's not like we would have uncovered something new. And I most of those things as the warm up, the things I wanted to ask about, where things like won their Young people in government anymore, I can get a pointed, I don't know, thirty one, thirty two yeah.

Um J F K, I think was what I was twenty seven twenty nine when he first became a senator late twenty late twenty um the hover dam was built by twenty five year old Young G I. I asked her in the in the podcast and um I think I cut a little bit of art because as a type question I were asked I think is super interesting, right? The other question I want to ask was and SHE, I had to answer a one millennial, I mean about generational change and other stuff.

But like I would have loved to keep dig on that. I would have loved to keep digging, and I don't have a chance to ask these questions, right? I think they're interesting like how do you run a government agency? Like what's important? What do you think about? What do you consider? What would you do differently next time? And so to me, the really interesting stuff is a stuff that SHE and others like her uniquely positioned to provide insight on that just nobody outs, right? Because everybody keeps asking about A N A. So that's the stuff I wanted to get to.

And if I have have the chance to talk with her more for in favor CER. No, David runyan, which what they do is they take the people who may break through them, they are Young, and then they let them help choose a new Young people who should get a lot of money, and by all accounts, like a hundred times more effective per dollar thing.

What the nh does these areas, and I know you to say that it's like, it's like sacred car and allowed to a attacking government. The nh, as you know, is packing more and more and more older, older people. And I think that means it's just I think it's tied to, unfortunately, not having any anybody where they're our labs about forty, fifty six year old that the desert be packed, their amazing labs. But I think we need to have a little, little bit more of the money going towards people on their twenty and IT seems like maybe government in general, I think IT becomes more institutionalize ed, more credential.

This is like more careful. Think two things going on. By the way, the first set of the nh was in his late twenties. amazing. One .

hundred, two hundred, three.

Like at this point of what wide society is that change? yeah. And actually, if you look at started under the same thing has happened, right? Patra collison as a good point of like we're all the Young started founders and there .

there some again and ung one right generation ally, go back ten .

years and IT was patric and IT was ock and IT was dealing from fig ma. You can kind of come up with the list today. There is many fewer, and I am starting to see some of them. And I emerged again.

right? Which I think is .

interesting gotv. I guess pretty scotland, pretty Young. But it's kind of like where are the people who have very clear outside success really early? Those are the people that do amazing things for twenty years after that, right? That's actually a platform then to have enormous is .

very twenty one and have a lot of other things. Yeah exactly right.

So where are those people?

And seeing a lot more of my favorite companies built by people who are second, third, fourth time entrepreneurs.

Are you seeing that too? I'm seeing A A bunch of that um and that's almost a negative sign. Nothing in that case, but we are all the Young people for the first time who are really succeeding. And it's not about starting, it's about succeeding.

And I think those are different .

cultural. And like one theory could be it's a cultural shift where we have so much helicopter parenting that these people are the last brazilian and or they don't take rest. You can make an argument around how something there's all these jobs, a big tech that are incredibly lucrative when you're really Young. And so that distracts you. IT would have been tough as so on.

Off for me, half a million dollars a year. I was twenty one. I I wouldn't gone to start pound out, to be honest.

and so that maybe you would have ended up at google or matter .

for a few years or something, then put you .

on a very different trajectory. Think about the world different, right? So this really interesting question of missing Young people society. And I just I don't know what the driver is. And again, Patrick think has brought this up in a really Chris boy and he's Bobby thought .

about that much more with progress in important ways to yeah .

great with progress some ways. And again, I think it's about this art of impact that some people have over their lifetime. Like, look what musk has done.

But his first successful is zip to in the nineties, right? And that money then went into x stock hom such paypal. And then that money went in to rockets .

and cars and you know a compounds.

And but so it's it's stuff like that, right? And so um and you see these other x right? Mark and rison was a fee up right with nescafe and he was on the cover of time magazines is twenty two twenty three yeah and then he had two decades as to do more stuff. And so the question is, you need those early successes, I think, to have massive social change because then you can stick around long enough.

No, it's ue. It's sure a lot impact we were having else because we were able to do suffer. I will talk about A I with you.

I know it's like the most trite thing been involved. I guess the machine learning in general AI, since it's very, very early on doing A I google early on. Um first of all, like give us some perspective on like the investment in here.

And i'm seeing i'm still seeing like lots of make rounds, a hiked bubble. I heard the cost of each one hundreds or tuners whatever coming coming way down right now. So like like what's what's going on.

We and I mean A I is in my pink dramatically under hype right now. And I I think part of that is number one, rising massive actual revenue and impact without that much adoption, right? So um assures last quarter that did like twenty eight billion dollars in the quarter.

I think this is ten fifteen percent of that lift or growth was from AI products without be one, two and a hf million or something. A man, right? amazing.

But that's the infrastructure people are building on. We are paying them. So are they making people? Are the things on top of profitable? Or those things just .

raise the money and build on top of are doing right. My journey is rumor to be doing extremely well financially. There's a variety companies that are scaling very rapidly, but there is a very early days, right? ChatGPT came out lesson two years ago, and that was a starting gun for a lot of people .

to really GPT three was only two years.

amaze. I chat, G P T. G P T four came out eighteen months ago. GPT three came out and covers twenty one, just like about three years and three years. yeah. And three, three, you could can see things kind of working, but three and a half four when you really saw this, he change in terms of capabilities and charge P T S of that the sort of a post training thing um so on the one hand I think is ramadan understand because very little adoption is still creating these massive um revenue streams but also corina. The ftc company mentioned that they reduced uh their customer success in by seven hundred people yeah by adopting A I and they simply had twenty four seven availability and thirty languages with the higher etretat score, faster response time, higher customer success we've seen already.

Like a lot more money going into this way than anything else in a very long time. Use how much money is what would be a not only .

helped yeah I think what people are really misunderstanding is what is gonna the end product. And I think the end product is units of cognition, right? You're onna be effectively paying for a renting um something that's gonna think or do things on your behalf.

And that could be legal documents, that could be a customer success team. Effectively IT could be a soft ig engineering team, which is basically the odds reading good for you. And so really, this is a revolution in terms of units of cognition. And I think that's a very under thought about is actually what the product is.

is anything is the pulling out right now, for example, in the service area, million about the companies area, it's not sure cognition. It's like right now, it's like it's doing things to save people tons of time and make each person able to do like three times or four times as much.

So so like for example, like an or like you can bring up right away, this ten screen are most likely look at all around and that way they go right away, they see in that way they save them just like five minutes each time. So say, i'm like it's making three times so that so is interest because right now that's more machine in bio sis. Do you see that? Yeah just staying this machine and BIOS .

is so do you see the older actually preta couple years ago, I read this blog post I never published. I just never got around to polishing.

I just put out, I rote this maybe five, six years ago, which is basically positing that the'd be like three errors of mankind in terms of, or three areas of intelligence, right? And the first era is basically people in the second era, some hybrid era, where is a, uh, you know, if you look at the number of intelligence units are what everyone to call IT the relative brain power, you'd have a mix of, you know, human kind of machines. And then eventually it's gonna mainly dominated by machines that turns a sheer number of brain equivalence .

that all exit there. This is a bit like some more .

you can scale IT fast with with a bunch of chips and software. But the ideas .

these are like serving us and they are like work for us. Or as the idea that some point they like.

they're I don't know what the very long run looks like, but in the short run um you my team actually looked at this. If if you look at, for example, the this is world in the U S. Uh if you look at us and enter Price support in the U S.

Is about have twenty dollars a year. Um if you look at the areas of a ee are the areas of services that AI can transform, it's probably five trillion dollars in headcount, right? Cos it's just paying employees .

was actually all services in the us. Our best estimates about forty percent that right now could be transformed .

A I you cut of that and you take ten to twenty percent of that. But our math was basically, if you say take a ten percent cut and transform, an employee had count costs. In sas software, you're back to have a trillion and you've recreated the entire enterprise software market cap. No, I think in the two point .

one in trillion and that we think could be addressed right now, stuff like health care billion and just six billing and whatever legal some legal stuff, we think already you can double a proclivity, which is kind of a trillion.

which which is huge. yes. And then you capture .

some upset of that, right?

So that we again, the unit of work that we're selling is different from we're selling event should be selling the units of productive intelligence. And that's a new type of skill that nobody has ever thought, thought about or had before because the prior waves, a machine learning, I think, where people misunderstanding all right now as we've been talking about ee for twenty years.

And that's because we had these prior waves, machine learning and convolution learning networks and you know arnes and gas and all these other types of networks um but they were a very different fundamental, both architecture but also type of product. Really, what those things used to be good at or still are good at is pulling out statistical associations between large data sets. And so they are really good at at stats is basically that pier wave, the current wave is really powerful and that it's actually understanding and generating language and images and other things associated with that is basically um and this is based on this new architecture called the transformer architecture, which was invented in google in twenty seventeen.

That's been the big transformation. So it's actually a different technology curve from what we had before, but people call about them ai. But I think we need almost like a segmentation of these things because we're on a .

different trajectory. When you say curve like this, to me IT seems like curve more like this, but I could be totally wrong. You actually you think it's going to I mean, it's going to be a GPT five and six or whatever from someone else to it's going to just be expensive.

Better than what we have, right? It's not only expense cally Better, it's impacting other fields. So you look at what amo is doing for self driving or total sa driving. We've moved over .

to a transformer backside and some they see dramatic improvements. And I mean, elan rid all S I got go and and a ee.

instead of having all these little cases hard coded in. And that's happening in .

every single field, I want to push back because I agree it's like dramatically affecting all these fields. I agree, like I said, we're using up for services already tripling the proclivity in many cases. But that stuff you could do with like we l lambs are today.

sure. So the question is we go like this. The question is, like R, L, ms, going to exponentially keep getting Better to the point that they, like totally changes, improve that? Or are they hitting some kind of local maxims?

Asm, as far as I can tell, a lot of researchers in the field, the way I think about IT is almost like A L M steps, right? And when you had GPT three point five and you went to four, or suddenly legal, open up as a vertical called hard.

and they showed me side.

right, you had a model to take something you could do legal. What's the next step? pocatello? I think that when you could do with like GPT one, you know, you have these steps.

And so this latter that you're climbing and adventure, you going to hit a point where can do all of human services, right? And quiet. Or eight or or we're stuffer .

are building now to do these services and win with those companies. Still be available in five years because they can replace IT with A M A I S goes long is going to be some new .

is he going to opening and itself that just OS all of those things. Law firms and I asked them, how do you think about this? Some law firms, traditionally some of the worst software buyers in the world, right, they don't really innovate, are locked down because the security and privacy in other very legible ate reasons um but they tend be very bad buyers. But they were adopting harvey really quickly.

And so I asked them like, how do you view the future of A I? What do you think is gonna en? And I was surprised by how thoughtful and forward thinking they were on this, where they had, look, we think, the nature of a affirm al change because right now we hire and making IT up fifty associate every year, and five of them will make IT a partner.

But we now think that in a couple years, because just hire five assists to begin with, then who becomes our partners? What do they learn along the way? How do we teach them? How do we scream them? How do we ever a portfolio of people so that some work out in, some down? yeah. And so they're thinking about the legacy of their law firms and like we don't know how to think about this future world, but IT also means maybe have one senior partner and two social and thirty bots yeah you know and so the the whole thing kind .

of shifts is also something. So with harvey, you at the model, we are science of law from a lot of my companies when we call AI services, we're literally replacing like the healthcare billing charm like like we're in. We're doing at all ourselves, but we're doing IT way Better for action. The costs. Like are you seeing .

a lot of those companies right now?

And I guess what's your your intuition? Let's say, let's see if they take your argument to scare people that that you're right and it's gonna go way Better or GPT five, six, seven or the equivalent. Is IT the case that by GPT seven, maybe they just like do IT out their company themselves even Better than our source? Is company always their competing?

It's possible, I think, IT comes down. What is the other tooling is needed around what you're doing or what you're providing? And so the things that i'm bullsh on or or you on a workflow for a vertical, yes.

And so everybody, that vertical is that work direction ah. And so if you upgrade the model, I just makes that worker Better. So doesn't matter what I see.

I will be chAllenges model for me to briefly, I talk about five layers for the airport sucks. The bottom layer is like invidia, gyps and everything. The next ones the data center is over friend's family offices bring comes money into.

And then it's like the models, whether it's elan's x or anothah pic, then it's like little of four would be the tools applying A I and the number of five is actually only in the workplace services company or this is that's why i'm building like probably zen things right now. Uh if you're doing a lot to um so so so which of those five are you most polished? Which are you spending time on you?

Yeah I mainly spending time on the .

top three and talk about IT.

I I have a slide that literally has sort of uh presentations I give in stuff. The bottom side of the chip side actually invested in two companies seven, eight years ago thinking that there .

would be in video competitors and I told .

just so good um there are there are just very good and there is still room for other players. I'm just saying like they really executed well. But yeah, now i've mainly been focused on the top three of those layers. The other thing i've been doing is actually been backing i've now back to um A I driven biot where the idea is you buy the asset itself and you can radially change the cost structure increase of leverage of our organization um by putting an A I in a deeper way and then that entity can go and roll up other companies in its vertical.

I think that's very nice, is what we're doing to like with the health care building in the logistics building, other ones like that will grow gani ally at first.

But orginally and these principles for an organic growth with A I that are pretty crazy er I know this is buying um call more things what we have you on A I A and know you have children as as do I um the world's going to probably very different in ten or twenty years like like what are you paying attention to for them? What do we do differently with education with this effect rates? Our kids?

Yeah no, I wonder about that a lot and I don't have look a answers and i've asked some of the world's top researchers about this right because I talk I talk to many of them with some regularity and I am like what like watch in my king study you know what you what's a good thing to know in the future um so that that you know the positive of the sai way for this kind of stuff is that eventually each person will have a custom tutor which is helping them learn really deeply at their own pace right? And I think a eyes is perfectly suit for that is going me very exciting world where your A I really going deep with your kid on different topics. And there S A research from shows kids that received one of monitoring um learn dramatically faster the .

logs into the great framework.

IT really is right. I mean, a lot of things that happened thousands of years ago turns out make a lot of sense and teaching kids, uh one on one um increases or performance by I think I covers one or two other as of magnus de right um I mean now one or two standard .

deviations S I A A lot of great minds of the revolution did get private or or something that we should be doing more of today, which is kind of I think his back and I was very strategic now and any .

make democratizing we should be able to get up because they should be so machine learning system running in the background and A I system turning the background. So I think that part of is extremely exciting. You know the other question i've been thinking about is just, um is there anything that you know if you can afford that you can buy for your kids now? So that is a durable in to the future, because if you imagine that A I is gonna bend a lot of industries and a lot hipper jobs and is IT like land, that is just like what's what's actually worth something in the future?

You have girls fun. Come on.

So it's crazy to think like what's valuable twenty years from now.

we do have .

a lot of land.

Austin, I ad top companies that are tracking talent, right? I don't know. I feel like I don't know. I feel like companies with a great talent that are good to the I has .

to be no predictive value of what companies will exist in twenty years. Maybe railroads, right? You say, okay, I the railroads are basically localized monopoly, right? They own the tracks. You're not going to build up the norm. But agree yeah no, there's a stuff like that is really s so .

maybe by raillard stocks just go on amErica american optimistic but but nothing. Now to teach kids in particular actually is anything particular?

It's math, computer science, resilience, a physical fitness like I feel like it's kind of that stem that same you know how to write, how to think um how to express one self creatively .

philosophy maybe a little bit .

so in a bit yeah depending on which part a lot of physical hy out there I S .

liberty in order in all staff yeah it's like .

what is the moral fabric and framework that you want your kids to be part of and then also I feel that some form of um religious upbringing as useful, if only to protect them .

from vague religions sort of coming N N .

against part of everyone.

Can you do you think the traditions really important to have that that's important. So you know you know, obviously really important work on A I and biotech on the frontiers when I think you are passionate about.

I believe, is building inspiring new monuments and starting and i'm looking to bring on somebody to help me drive. I I if you look at every society at its peak, or at least on its way up, you'd always have this large scale inspiring monument.

Celebrate wins, and you celebrate spe the future.

inspire the next generation of people like watching a space rocket launch, right?

It's so inspiring. My friend rodney cook in georgia is the head of the national monument foundation, and we have helped build one really important is something. That he's more in a class architect, which is like, I love that style and is a lot of beauty and truth them up.

But IT is is something we do want to do IT in traditional ways. We do want to do IT in new ways or both. How do you think about that a little bit?

I mean, if you look at a lot of the really important so there's obviously the seven wonders, the ancient world and all the stuff. But also more and more recent times, you look at things like the statue of liberty and people coming in the Alice island, a sort of a major entry point for immigration, and you'd see this inspiring statue, right, the inscriptions and everything else around that, that that helps motivate people to do amazing things, right?

Are you look at the iphone tower, and that was built for the world fair that was held in paris in the late eighteen hundreds, and IT was impossible. Be a testament to french steel making, right? And so that was postal technology progress. That was how we can .

do up with steel. Isn't that amazing? You're showing, if do you believe in these things like the golden ratio and like ideas of beauty and stuff.

have lost some form of sense of beauty society. We don't aspire to make giant beautiful artifacts.

I think our culture rejects IT as well as A A moxy.

And mark said in the question is, why and why would you mock something that inspiring and creates a sense of light, a sense of hope, a sense of purpose um and I think that large scale monuments ments to do that right? And so um and that's true in every society throughout history .

why I love that monumentally doing a lot of things around like neoclassical and the tricia as well as trying to be inspiring. I am told with you on that we actually start the american optimists trying to push back all in our country innovation. Our terms about comes ten or twenty years with with .

these positive so much exciting stuff coming um um by the way, one could argue that the people of so even stuff like that right could be really um if you think about how do you make that a public art work or something else, this buildings far more people and in terms of the future, I mean, obviously there's so much coming that can be incredibly positive for the world. Um there was sort of stuff in the eyes, stuff in health, stuff in education.

Um there's a lot of different um areas that we can transform society in a really positive way. I think fundamentally, um we are to talk about education side and how I think I I actually gonna have huge impact there. I think from a health equity perspective, A I can be transformed there. Um google released a model I think two years ago called md palm two where um if you compare the answers from that model relative to physician experts and not perform them words, I use physicians to train the model.

The model will get work decision are way behind, are way behind. Now what you can do AI, they don't seem to know what.

I think we can be added physicians where they can be all for them. But also imagine if anywhere in the world, if you had a as you know, billions of people to have smart phones or covalent ts, if you could take a photo of something upload to add some tax to speak into IT um and then get out um uh a medical answer that the change is actionable and it's equivalent of stanford medical care nb Anderson and or whatever ah that's incredibly empowered .

for the world right? If anybody .

anywhere they do that working on um what's train, who's who's on my team is um we're we're onna take um a thousand a great works that are off copyright and we're going to translate them into one hundred languages using A I and we're going to do an audio book uh for all of them N A I and they are actually onna build a module where you can interact with and chat with the book and away where has the intelligence of the full foundation model but it's ort of represents the bulk persona.

And so um you know we think that is almost like a modern library, Alexander, because if you look at throughout history, public libraries have been massive public goods. And there's two threads of that. One is how do you create access for anywhere in the anyone in the world through any modality? They should be able to hear IT. They can read IT whatever IT is um they can interact with that um to gain new knowledge right help summarize this chapter for me and let's talk about .

IT for I just think it's really exciting.

That's one thing. And IT could be religious works at the bug guitar, the bible or whatever. Could be philosophical treaties that could be works of mathematics that could be great books, shakespeare at a and so we think there's a broad swath of stuff that you know we can incorporate part of IT. So um that's another example of where I just think we have these amazing ways of technology coming and they can be used and osprey to visit. People aren't doing IT, and they could be incredibly powerful and empower ing.

So as a lot to do, health, art, inspiration. And we we we have a lot, lot of in.