E122: Is AI the next great computing platform? ChatGPT vs. Google, containing AGI & RESTRICT Act
01:24:29 Share

Hello, jack, hell's here. Hello jack, Harry and we show up.

I ve been heard the whole time. I I was just, uh I was just having some of these beautiful shorted, rusted satis and we promise when I went to the store, I kid you not, there is a shelf on these all flavors s available .

except .

one flavor vinegary re.

we move the market. We move the market.

I am not kidding. I go to the fancy, you spoke the is and truck. I went to the rallies and trucks. The are and they have, you know.

all this over private is called article.

That's what I said, the art stuff, the artistic food, the artisan brow. When I had this, I kid you not, spicy soup y no salt, every shelf packed. Then there's one shelf I can see straight through to the ice cream.

but not see.

And I look at the tiny little sign, salt and vinegar shelled nuts c south, he sought in vinegar to mars. Shelled nuts sold out across the country. You know, I cannot .

recommend these more highly. They're incredible. You can recommend your salty nuts work. They are delicious. My salty nuts are delicious.

man.

We source to the fans .

and .

they got. Do you see joe mansions high heater opel in the wall street journal of, oh my god.

yep.

je mansion went for IT with joe mansion running for president?

He is, I think, okay, so let me have sex right there. sex? Joo, nicki Haley and who's the guy .

from floria?

By the way, there is a big defection that was leak this week. Ron latter flip from trump to to scientist. That's a big one because latter is good for a lot of money. Five to ten million .

in this to match. What impact have coming to the west? I M. He was talking .

about the insane ery of the Better administration .

to control costs and how everybody was in component and IT. Certainly, there are some ways we can control some spending. Everybody needs to grow up and get in a room and just managed the budget for the american people in stop playing power.

Yeah, I think the has line of the article actually, to your point, ja was much worse than the substance of the article sacks. But if you see the headline, and if you can just throw IT up there, he was brutal. The headline in the byline of the article, I think, was more damaging than the substance of the article.

Brita's inflation reduction acts betrayal. Instead of implementing the law as intended, his administration .

converts IT for ideology.

Sil, he permission to a private right.

And by the way, I think if you guys remember, we talked about this when that act was first published. And if you guys remember, right, I think I pulled up the C P O data. The C P O model and IT showed for the first five years.

This thing burns a couple hundred billion dollars. And then there are some expectation that will be some sudden boom revenue in the out years and then you make the money back in the out years. So it's total like accounting and engaging for him to have made the claim in the first place that the R R A was actually gonna like a net deficit reduction or that reduction. In fact, it's all just accounting senate against and and if the massive spend package, particularly the new or term when IT matters most.

I think I told you guys this, but I think this is like, when is the last time I was in washington? Probably what is IT march now? So maybe was january.

I was there and I saw human and mark Warner and I spent about two hours with mention. He is really impressive. He's cool.

He's interesting. He thought for his moderate mentioned like a formidable guy. So this will be really interesting if he steps in there and you write your check. I've probably right a to both to be once .

feels like a good ticket to me. I ve always .

wanted to see the ross demotic republic and merging .

somehow.

and like running together for David freeburg may have just come up with one of the most disruptive ideas in american politics has ever .

been floated. Oh h my this to my comment on this.

So first, all I member when you mention did a good job stopping bite in three ha trillion dollar build back Better remember IT with him and cinema that were the hole matching compromised and gave by in a seven hundred and fifty billion dollar version of IT. And I guess now is complaining that bitten and live up to his end of the bargain and doing the defter reduction.

But quite Frankly, many commentators said at the time that the bills claims deficit reduction were proposed ous, and that would never happened. So quite Frankly, you mentioned shindand ugt or put linked by biden. Everyone was basic saying they ll never be any default reduction out of this bill is just more spending.

So I don't really feel bad for mansion here are saying that somehow he is betrayed by e bye. You should know Better now in terms of him running yeah I think as a democrat who's figured out how to get himself elected in the west Virginia, which is a plus twenty red state, he obviously knows how to appeal to the center. The problem for ham is just how do you get the democratic ty nomination because he's far to the right of your average democratic party voter if he wants to run as an independent.

That's a different story and that would really throw a curve ball into the race. But I don't see him doing that. I think it's kind of stretch. And this is the problem with a lot of these fancy candidates is that, you know, centuries, or moderate voters might like them, but they can get an omission. Their party.

You like trump at obama, those were .

fancifully .

not a fani ult.

He was outside, but he appears the base of the party. He appealed to base the party. What i'm saying is order to get the ominous of a major party, you have to appeal to its base.

And I don't think mansion appeals to base in my Carry party. He's out of step with IT. He's out of step with IT in ways that I like. Don't get me wrong, but I just I don't see how he will get an ominous ris. Christine.

what do you think of him? He seems like he's about to come in the rest to David. I win, right?

Listen, everybody.

Welcome to the all in pocket.

It's like .

episode one hundred, something with me again today. The rain man himself, yeah, David tax is here. freedoms.

g. Is in his garden at his home in paris. Spring has bunk, the queen of kinda.

And of course, the dictator himself chmagh. yo. Pedia the silver fox. Look at that little tough of silver hair, so distinguished.

I got a haircut from somebody recently who said that people go to her and ask her to put the silver thing in their hair.

Really.

I don't have to worry about that. Looks like he's in smr village there. What is that backroom background .

that is a scene from.

嗯， OK.

I'd like most of my backgrounds. I know my the mood in the moment of the week.

And you guys just totally, totally denied half the beta males in the youtube comments from being able to guess the background.

Thanks a lot. Reverse max than I used a ChatGPT g .

tomorrow figure.

I don't know was this week now.

Okay, well, let's get started. Come.

let's get started. okay.

Thank you.

Welcome to the .

world's .

greatest forecast. OpenAI launched a bunch of ChatGPT plugins and I don't know if you saw but David sax or a blog post with ChatGPT. It's an amazing back and forth.

I read this back and i'll explain what you did. Sax, this was really like one of the best conversations. I've seen a ChatGPT pop appear on the screen.

But well, any idea for a blog post about the use of a good tactic you could call gift to get? I thought IT would be an interesting tactic for AI starts to use if they're trying to get a hold of the prieta training data. So for example, if you want to create an architect AI, you need a lot of plans.

Or if you're going to create like a doctor AI, you need a lot of labor results or medical reports to train the AI on. And those are hard to get. OpenAI doesn't ussa have them yet.

So there is an opportunity, I think, for startups to create these ais and different econ professional verticals. So the gift to get technique would be you give points to your users for upload that data and then they can spend those points by using the A I. And anyway, the company that came up with this gift to get tactic was a company called jig saw almost twenty years ago.

No one remembers this company. I'm kind of dating myself because I remembered IT, but I decided this idea. G, I wonder if the gisar approach could be used for A, I started, so I started by going into ChatGPT, and I said, hey, have you heard of jigsaw? And then I had, and then I said, tell me about it's gift to get, you know, approach.

And then I said, with this approach, work for a eye star of that one proprietory training data sets. And I said, yes, this is a good idea. And then I gave the architect example.

And I said, can you give me more examples like this? And I gave me, like, twenty more examples. And then I asked IT just to flesh out various kinds of details.

I went down some cold disease that and use. And then at the end I said, can you summarize everything? We just talked about a blog post, and they gave me the first raft of a blog post.

I then did a substantial ounce of editing on most of the blog post, although so that I just use rebeca. And then I had a couple of people on my firm look at IT. They made some good suggestions.

So it's not like the humans completely out of the and then I copy th Epace o f m y e dited v ersion b ack i nto C hatGPT, said, here's my edit. And then I asked for some suggestions and made a few small edits, and I said, okay, great to incorporate the edits yourself, gave me that final output. And then I posted on sub stack a blog that probably would have taken me a week to research.

And right, if I done at all, I was able to do in a day and I can't see myself going back. Now I think this is just how you to write. All my blog post is is use ChatGPT as my researcher, as a writing partner, some cases in an editor. But i'm definitely going to run this .

through the thing that I was struck by was just how kind and generous and thought for this conversation was. And I just thought I ve never seen sex. Have a conversation where you were so kind to the other person and thought for right about now all your friends and family, like how do we get sex to have this conversation with us? You were super .

kind to the A I, because it's not a person. IT was a robot.

But just in case that takes over the world. J out, you can be too careful. But no, I think listen is important to give the A, I.

so kind of perfect.

perfect. For example, A, I actually gave .

me some information about jigsaw point system. Again, the rewards that they used. Yeah, and I was just in text, so I set down below. Hey, can you spend that? I was a table and I did .

instantly. I incredible.

I made exhibit.

yeah. I said, thank you. That was like, delighted to you. I mean, this is, this is a road. This is a literal road. You being a community, being in all the money that you've spent on therapy and just trying coaching to be nice to people, you're just nice nature.

Way to this is me and I know why you're sitting. I heve it's fine. You have it's much sax saying, thank you to the A I perfect.

This is confirmatory what we knows. David wants to live in a set of highly transactional relationships, ideally with a machine. Who can then eventually help me making money?

Can I ask question of the city sex? Did you what did you enjoy more? Working with your team of humans on this or working with ChatGPT? Which one is more enjoyable for you just personally? Well.

I think they both were. I would say that the human contributions were essential. So because about enjoyment, know it's about this. This is a job to get done. But but I definitely spent things up enormously.

I personally find the hardest part writing a blog is when you're staring at that black eea paper and just having to like spit out the first thousand words. Yes, it's a so time consuming to do that. But again, if you start with the first draft, even if it's not very good, then you can just .

edit IT and it's species one.

Yes, yeah, it's great, important.

Yeah, I actually trusted IT.

I I know that you probably should fetch out anyway because I can have luCindy, but the things that were saying made so much sense to me that I didn't think he was late.

Well, this is a great moment to be into what OpenAI did with plugins. These came fast and furious this week in a bunch of folks who had, you know, started vertical zed ChatGPT based project M V P is really like, oh, maybe my project M V P is now dead because inter card, open table shop, ify slags, aper examiner, obviously, like, if, then this, that matt is a very wide ranging tool that allows you to connect A P S. From a multitude of sources.

And what is all what you do at the end of the day, is have ChatGPT paying one of these sources, just like and APP my do or some custom software might do pink the API and return data. So hey, what tables are open on? Open table? Maybe shop I fine me, uh, things to buy in this category.

Exeter up. And so people have started building little scripts used to call. These are when magic leap was out of internet agents and the concept of a software agent that's existed for a long time, actually, in computer science.

I'm sure free brick will give us some examples of that. But also ChatGPT can now use a browser. So that means you get around the dated nature of the content in the corpus. Somebody did things like, hey, build me a meal plan, book me a reservation for friday night in open table, source other ingredients and buy a SATA night on court, and then use something like, well, from alpha to, you know, calculate the calories, except as, so when you saw all this drop sacks, what did you think in terms of the opportunity for startups and to build these intelligent agents, things that will do if then, if this, then that, or just background tasks over time and you can actually leave them running? Yeah.

I mean, I think this is the most important developer platform since the iphone in the launch of IOS in the APP store. And I would argue maybe ever in our industry certainly sense the begin of the internet.

I think there was a question when ChatGPT launch on november thirty and people will start playing within december, what exactly open in a product strait was going to be? Was this just like a proof concept or a demo? And they even kind of call IT like a demo.

And initially, IT looked like what their business ma was. Gonna was providing an intelligence API that other websites, other applications could incorporate. And we saw some really cool demos like that notion demo of other applications incorporating A I capabilities.

And so initially, IT looked like what OpenAI was going to be was more like stripe, where in the same way that strike made payment functionality available very easily through development platform, they were going to make AI capabilities available through their developers platform. And then I think a funny thing happened on the way to this announcement, which is they became the fastest growing application of all time, talking about ChatGPT over one hundred million users in two months. Nobody else has ever done that before.

I think he took the iphone beyond two years, plus gmail, google, those products all took, I think, well over a year. So this became the fastest growing sight of all time. And I think with plugins, what they are indicating is that they will become a destination site.

This is not just a developer platform. This is a destination site. And through plugins, they are now incorporating the ability to basically, you know, anything you could do through an application you'll now be able to do through a plug in. You will tell ChatGPT what you want done. If you say, hey, book me A A plane ticket on this state, IT will go into kyats plugin and do that you say.

book me a plane ticket than an for of I and alexa realize because those we're very rigid. They had no intelligence, right? Free burger, if you, if you want a theory to do something specific like useless, or to go get you open table IT IT needed to be pretty specific and and IT didn't have any kind of natural language model behind IT. So this is taking existing as and putting a natural language layer in front of IT, which makes IT, uh, you don't perform a little more naturally. Is that what we're seeing in a for, I think.

provides access to a corpus of data and the sweet of services that are not well integrated into a search or chat interface anywhere today. So you know, knowing what restaurants have, what seats available is in a close service in, is in a data warehouse Operated by open table. And now what open table can do is provide an A P, I.

Internet data via interface. And we can allow ChatGPT to make a request to figure that data out, to give a response to a user where they can ultimately benefit from transacting and allowing a service. This closes the loop between search and commerce in a way that google cannot and does not do today.

And I think that what makes IT very powerful, we've seen this attempted in a number of important ways in the last couple of years with alex a and apple home and and google home kind of integration via the chat services that they offer. We speak to the device. But the deep and aggression that's possible now, and the natural language way that you can go from the request all the way through to the transaction is what makes this so extremely powerful.

And I think, you know, the points I made a few weeks ago, and we first talked about, you know, search, having so many searches that are done where the human computer interface presents a table or presents a chart or presents a shopping list in a matrix. That's what makes search such a defensible product, I think, could theoretically ally be completely obviated or destroyed with an interface like this, where you can write the ability for ChatGPT or whatever the the the core service services to actually present results in a table, in crates, in an interface, in a shopping list and actually close the transaction loop. It's really disruptive to things like commerce providers. It's really disruptive, you know, someone's commerce platforms. It's really disruptive to a lot of different industries, but also introduces a lot of real opportunity to build on top of that capability and that functionality to rewrite and ultimately make things easier and Better for consumers on the internet.

What do you think to math? You're looking at this. And IT seems to be moving at a very fast pace.

Over hundred million users. They put a business model on IT already twenty box a month. They have a secondary business model. Hey use the A P I will charge for usage and then you layer on WhatsApp.

Er and if this then that had already sort of established in the world, which is APP, but nobody ever really wanted to write script. So that seem to be the blocker you go in zipper. If this then that it's four, five percent of the of its people who want to custom stuff, people who want to tinker.

But this seems to now with a ChatGPT chat interface opening up to a lot of people. So is this super significant? Or is this a commodity product that, you know ten people will have over sitting here next year? On all, in episode two hundred and twenty.

I think you are asking the exact right question and you use the a great term, like in poker, if there three hearts on the board and you have the ease of hearts, you have what's called, but not blocker right, which means that nobody else, even if anybody else, has a flush, they never have the best fish.

And if flushes the best hand, there's a lot ways that you can manipulate the pot and eventually win the pot because you have that ease of hearts and nobody is. The concept of blocker, I think, is very important to understand here, which is what are the real blockers for this capability to not be broadly available. So think you have to segregate.

You have the end user destination, you have the language model, and then you have the third party services. And so if you ask the question, what is the incentive of the third party service? Well, the shareholders of.

A travel site, right? They're not interested in doing an exclusive deal with any distribution and point. They want their services integrated as broadly as possible, right?

So I think the the answer for the service providers is just like they build an APP or IOS. And for google, Allen, you know, if they could have justified that they would build an APP for a gaming council. They can.

They should. They would. They do, right? So that's gonna get commoditize and Brown available. I think on the L M side, I think we've talked about this. Everybody's converging on each other. In fact, there is an interesting article double released that said that there was a handful of google engineers that quit because apparently barred was actually learning on top of tragedy ity, which they felt was illegal or unethical or something.

right? So, so the point is, I think we've talked about this for all, but all of these models will converge in the absence of highly unique data, right? What i've been calling these White truffle s so if you can hold White travels, your model will be Better.

Otherwise your model will be the same as everybody else is model. And then you have the distribution and points of which there are many whose economic and sentence are very high, right? So facebook doesn't want to just sit around and have all this traffic to ChatGPT.

They want to be able to enable instagram users and what APP users and facebook users to interact through messenger. What have you? Obviously, google has, you know, many hundreds of billions of reasons to defend their territory.

So I think all of this, to me, just means that these are really important use cases. As an investor, I think it's important to just stay a little patient because it's not clear to me that there are any natural blockers. But I do think that dave is right that it's demonstrating the use case that's important.

But it's still so early. We are six weeks in. Yeah, I tell I think .

there's a couple of great blockers here or there's going to be an eminent banana for silicon valley. If you look at certain data sets, read IT stack overflow for programing and cora, these things are going to be worth a fortune and to be able to buy those or get exclusive licenses to those, if you're maybe google board or if your ChatGPT, that could be a major difference make or twitter's data set, obviously.

And then you look at certain tools like asier. And if the theyve spent a decade building the sort of, you know, ma ta A P I, that would be an incredible blocker. I think this is going to be like a ization. Many .

exactly .

say, I don't think these are not blockers. I don't think this is the ease of hearts on a flush board. I don't think so. I think that these things are really interesting assets. They are definitely trouffle in nature, but they may not be the you know ten pound White travel from although that we're looking for.

But the M A I I do you think this would be like incredible now.

But the only reason I say that, again, is IT is just so early. Like I in the text, I mention this, two guys I remember and sex and I were in middle. This, we were both right at the beginning of social networking.

Sax started gene. I was in the middle of aim when all of a sudden we saw read started social net. Then we saw friends are gets started. Then we saw my space get started. And you have to remember when you look back now, twenty years later, the winner was the seventh company, which was facebook, not the first, not the second. IT was the seventh, which started two and a half years properly, after the entire web point to a phenomenon.

started m with search, by the way, where google was probably to .

you to be a real student to business history. I'll just say something that's more matter, which is if there is something that i've learned on the heels of this S B B fiasco is that there is an enormous amount of negative perception of silicon valley and Frankly, a lot of disdain for VS and prognosticating technologists, right? And I think that we .

have to be cast.

I think we have to be very careful. Yeah and and I do think that we are an example of that because we are the bright shiny object of the people that were successful. And the broad makeup of amErica thinks that we're not nearly as smart.

We all think we are. And after all of this, money has been burned in crypto land and N, F, T. And all of this web three nonsense to yet again whip up the next hyped cycle, I think doesn't service as well.

So I do think there's something very important here. But I think if we want to maintain reputational capital through this cycle because government will get involved much faster in this cycle, I think it's important to just be methodical, thoughtful iterate experiment, but it's too early to call IT. I guess what I would say.

yes, it's definitely all to call IT. But sax, you're saying explicit, you think this is bigger than the internet itself, bigger than mobile as a platform shift.

It's definitely top three. And I think that might be the biggest ever. I think look, I think things can certainly play out the way that chemotherapy ying.

However, I actually think that OpenAI has demonstrated now with these platform features that IT has a lead, a substantial lead. And I actually think that lead is likely to grow in the next year. And let me tell you why I think it's got a couple of assets here that are hard to replicate.

The number one, user attention. I think they've now got, I would guess, hundreds of millions of users and this thing is caught on like wildfire, must have been beyond their world to stream. I think that even surprised them how much this has taken off.

It's really captured the public imagination, and people are discovering new use cases for IT every day. If you are sort of the the number two or number three of the seventh large language model to basically get deployed, bind a chatbot. I just don't think you're going to get that kind of distribution because the novelty factor will have worn off and people will have already kind of learned to use ChatGPT.

So number one is one one hundred and seven as the eyeballs. Number two is with this developed platform, I think we should describe a couple of other features of IT. One of the problems with ChatGPT, if you've used IT, is at the training data ends in twenty twenty one.

And so you vary rapidly for many questions. Get to a stopping point where IT says, like I don't I don't know the end of that because I don't have any information about the last two years. Well, one of the pluggin s that open a eyes and surface called the browsing plug in and IT allows ChatGPT to go search the internet and not just run internet searches, but to run an internet search as if they were human.

So you ask, you ask, chat, P T A question. And IT goes to find IT runs a surge and then IT scousers through the list of twenty links. And IT doesn't stop until he finds a good answer, and then IT comes back to you with just the answer.

So that actually saves you the time of clicking through all those loops. And it'll give you the browser history to show you what I did. That's my blowing.

They also have a thing called a retrieval API, which allows developers to share preparatory knowledge bases with ChatGPT. So if you have a company knowledge base or some other kind of content, you can share with ChatGPT so that ChatGPT can be aware of that. And there are some privacy concerns, but the company is that they are going to sab oxi data protect IT.

As an example, i'm planning on writing a book on set using ChatGPT, and i'm going to put together all the previous articles and talks i've done as a database so I can then work with that in ChatGPT. So you're going to have more, more developers sharing information with ChatGPT. You're going to have ChatGPT able to update its training based on sort of the last two years, be able to search the internet.

And I think that as those hundreds of millions of users use the product and as developers keep sharing more, more of these data sets, the a can get smarter and smarter. And then was gona happen is both consumers and developers are going going to use or build on the smart A P. I.

yeah.

So this is well on self.

yeah. I think there might be I agree with much of what you're saying, but I do think somebody like facebook when they released their language model l which are about to is not going to allow ChatGPT have any access to the facebook corpus of data and then linked in will will do the same. We'll block any access to ChatGPT to their data.

And so then you might say, you know what, i'm doing something related to business and business contacts. I need to use the linked in one and they're just going to block other people who use the job to tell you, hey, you have to come to our interface and have a pro account, linton. And this all becomes little islands of data. And so .

I am not sure that .

I ter .

off the APP people .

who are smart state core. I doesn't let people use its data, so I just pick three. Those are three incredible data sets that don't allow people and cracked.

This doesn't. So people who are smart do not allow apps into their data. They keep IT for themselves.

I think there were a lot of people when the APP store rolled out that sore up and down there are never build a mobile up and want to give apple that kind of power that the internet was open, where as the APP store is closed and created by apple. And sure enough, they all, at the end of the day, had to rule out apps, even though, in the case of facebook, IT definitely has made them vulnerable because of downs am of apple.

I mean, apple now has enormous influence over facebook advertizing revenue because the users have to go through apple. They never had to do that before the internet. Nonetheless, facebook felt compelled to release a mobile APP because they know as extension al for them if they didn't. And I believe that was happening again.

I don't think this analogy, the right analogy, gy, would be google search does facebook does crag list allow their data to be index inside of goole search answers? No, right? They block that for a reason, and they will write to see.

to see this letter. So guys will stay out of IT, but look how much content google search ery has. And I think that ChatGPT will start by eating a substantial portion of search because again, you don't have to go through the twenty links that gives you the answer.

It's going to eat a substantial portion of browser usage and APP usage because you're just going to tell chat P. T, what you want to do. IT will go book your plane ticket. IT will go book your hotel room.

yes.

And the upset want to play in this, the I want to play in this will benefit. So theyll be a powerful incentive for applications to get advantage by participating. And me finish my point here. And then eventually they will be forced to do IT, not because they get an advantage, but because they're so competitive ly disadvantaged if they don't participate in that ecosystem.

I agree that theyll participate in IT, but here's the thing. What's going to happen is google gonna turn on boards. And I ve been playing with bard. IT is eight percent of ChatGPT already.

And then when they make bar a default, you know, a little snipped on your google searcher term page, or barter is built into youtube or chrome or android all the place, or they are gonna roll write over ChatGPT because they have billions of users already. So this advantage that you see today, I see that getting rolled real quick because you'll be on youtube and on the top right side will be barred. And when you do a search, it's going to say, here are other centage you could do, oh, you want to search mr.

Beast when he's helped people, or mr. Beast when he's given away more money or people who ve copied and been inspired by mister beest. All that can occur inside of youtube and ChatGPT is not going have access to the youtube corporation of data.

And then when you do a search, it's going to be the same thing. It's going to be on the right and side and it's going to be playing just like isn't being you turn on your android phone, they're going to make google assistant go right into bar. And google assistant is already used by hundreds of millions of people. So I think that google will roll. I think they're onna roll ChatGPT.

I don't know who's going to win, but i'm looking at this sexy more productively as a capitalist, which is what are people's incentives? Because that's what they'll do. Google incentive is to you serve ChatGPT usage by inserting something inside of their existing distribution channels to suppress the ability for you to want to go to the end notice building.

I think facebook has that same incentive. Oddly, even though microsoft is such A D partner, I think a certain asset of microsoft have that incentive. You're talking collectively about five or six trillion dollars of market cap.

Then then when you add in on amazon and theory an apple, what is incentive? I don't think their incentive is to let this happen. And I think if you look at the slack microsoft teams example of even a Better engineer product is excelled and widely deployed even at hundreds of millions of users.

Doesn't much matter when it's mock loverly distributed in Priced. And so though those things again, you may still be right, alarm thing is it's just so early to know and as slow and lumbering as some of these big companies are, they are not so stupid as to kill their own golden goose and or defended when threatened. So I think you just have to let, let IT see what happens.

I want to change to the point on google. Then we can move to the body thing, just make the counter argument, which is that I think called completely flat, flooded here, even though they shouldn't end, because they publish the original paper on transformers in two thousand seventeen. They should have seen more of this was going, but they didn't OpenAI use that paper and commercialized IT.

And the proof that is, there was just a lawsuit a couple of days ago, or at least a claim by a former employee of google who quit because he said that they were using ChatGPT to train their ai. So their AI is so far behind, they were violating the terms of use. All on, they were violating the terms of use of a OpenAI to train their own A I on ChatGPT. That's not a good sign. That's not a good sign.

I also hold on me.

I seeing the counter argument here. I mean, don't dismiss that out of the end. Give me a chance to explain IT.

Moreover, ChatGPT four, which was just released a few weeks ago, we know that OpenAI had that they were using an internally for seven months. So the state of the r is not what we're using, it's what OpenAI has internally. They're obviously working now on ChatGPT five.

And so if you're saying that bar is eighty percent of ChatGPT four, like I used for you is probably fifty percent or twenty percent of ChatGPT five. And who knows what the product road map is inside of OpenAI? I am sure that they've got two hundred ideas for things they could do to make IT Better and loading through.

But look, regardless, I think th Epace o f i nnovation e nvelopment i s g oing t o s peed u p m assively. I mean, there is going to be a flurry of activity. I agree, it's hard to know exactly how is going to play out. But I think this idea that it's a foregone conclusion, these big companies are just onna catch up with OpenAI. I think that there's .

a strong under argument. It's not a forgone conclusion where all the value will get captured just in any of major title waves. If you make the beats too early, you typically don't make all the money.

And IT tends to be the case. And IT has been in the past, at least with these transformative moves, it's sort of in the early third of the cycle is where the real opportunities to make the tons of money emerge. And there's a lot of folks that show you a path and then just don't necessarily capture the value.

I'm not saying that that's gonna the case here. All i'm saying is if history is a guide, all of these other big waves have shown that fact pattern. And so i'm very excited and i'm paying attention, but i'm just being circumspect with this idea that, you know, having been in the middle these couple of waves before IT, I made all the money by waiting a couple years.

I don't know that's going to be true this time around. So you could be .

right to your points act. I think it's clear and this is you know big ups to the opening item that they will be one of the top two or three players. Absolutely, we all agree on that, which is extraordinary. And himself and the top four players freeburg are obviously going to be microsoft opening eye will call that like whatever that little did I pairing, then google, facebook and then we haven't talked about apple. But obviously apple is not going to take this sitting down and hopefully you'll get in gear and have theory you make to the next level or they will just put out to pass.

Or if you were to look at those four and we're sitting here a year from now, who has the best product offering, who has the biggest user base? Just take a minute to think about that because you were at google and we all know the the world on the street is is the return of the king's Larry sergey are super engaged by all reports, every back channel. Everybody I talked to was saying that there every day they are obsessed with google's legacy now and making this happen. So what can you tell us in terms of who you think a year or two from now will have the biggest user base and be the most innovative among that cortex? Or maybe you think these other players who .

will emerge the vantage that open the eye has, which is the advantage that any call IT emerging you know, advantage competitor pass out is yeah outsider is that the incumbent are handy kept by the current scale. Much of the consideration said that google has had in deciding what features and tools to launch with respective A I over the last couple of years has been driven fundamental by a concern about public policy and public reaction.

And I know this from speaking to folks there that are close enough to to kind of indicate like google has been so targeted, has been such the point of attack by governments around the world with respected their scale and monopolist and monopolistic kind of behavior of some people who framed IT privacy concerns. You know, it's set at at the fines and the E. U.

Or extraordinary that so much of what goes on a google today is, can I get approval to do this? And so many people have felt so frustrated that they can actually unleash the tour kit that google has built. And so theyve been harnessing, focused on these internal capabilities.

I think I mention this in the past, but things like what's the right video to show on youtube to keep people engaged? What's the right ad to show to increase click through rates at at seta versus building great consumer products for fear of the backlash that would arise and governments coming down and then ultimately, ultimately attacking the the the revenue and the core revenue stream. And this is no different than any other kind of innovated flimm.

You know any other business of scale and any other industry historically ultimately gets disrupted because their job at that point is to protect their cash flow in the revenue stream and their baLance and their assets, not to disrupt themselves, especially as a public company, especially under the script and the watchful life of governments and regulators. So I think google has can aggregate probably good competitive talent, if not Better talent, than open eye and others. Google has arguably the best circus of data upon which to train the best capabilities, the best tool kit, the best hardware, lowest cost for running these sorts of models.

The law is, cost for serving them is set to a. So Frankly, their way behind the battle is there is to lose if they are willing to disrupt mselnet. This is the moment that Larry and surging should wheel those founders shares that they have, and they should wheel the comments that they wrote in that founders letter that they will always make the right decision for the long term for this company, even if that means taking a cost in the short term and disrupting themselves.

This is the moment to prove that those founder shares were worth, you know, that the negotiation to get there. And and I think that I was gonna require a real degree of scrutiny, a real degree of regulatory uncertainty, a real degree of chAllenge ed by governments and public policy people, and perhaps even a revenue hit in the near term to realize the opportunity. But I do think that they're Better equipped to win if they chose to.

Well said, well, really well said. I think the founder share inside is particularly interesting sex. The fact that I did nothing, I was the .

thing it's like, if they don't use IT now, what would IT take in when yet another, yet another case of the empty has no close, just a power grab by silicon valley exec, which was meaningless. Because if in this moment you don't real that power. And break that company into bits as you need to. What was the .

point of having IT if they need to come in and say we are going to give board results to ten percent of users and to have feedback who more square ies? Just one point I want to make that works. Who has more reinforcement learning than google? That search box is everywhere and people write question after A G I an. They so many people asking questions. And youtube might be the the transcript .

of and the comments under IT.

You know, the comments under the video, you have the transcript of what happened in this video, and then what was the question and answer underneath them.

Let me make the counter point, please, to my own point, like look at how girls now came after zc. So zc had his point of view. He strongly held belief that A R V R was the future of the platform.

That's what he wanted to bed into. That's what he wanted to lean into, is what he wanted to build the company against. He did IT, and then the financial analysts and the investors came at them and said, this is a waste of money.

Focus on making money. You have a responsibility. Shareholder of those founder shares. You don't deserve that telex voting right? Or whatever the frame might have been to get him to say, you know what, I act.

We are i'm giving IT up, and I think that we should also think about what's going to happen on the site. Google is a trillion plus dollar market gap company. Their their shares are owned by every public and downe public pension fund.

Institutional investor owns google in their portfolio, so the backlash against google making a hard bet like this and potentially destroying billions of dollars of cash flow in the process every year will not be easy to do that. The same sorts of letters that and obviously love girls. Now, you know, we can all defend him all day long at duck is what my may end up happening with with alphabet. They did chose to .

go as back sex. What do you think here about the founder share, specifically google chances of disrupting themselves and, you know, just putting this into every product and showing IT down user throats and catching up? Well.

I mean, with all, do respect larian, sergey, I mean, IT on the beach a long time. This reminds me, this is reminds me of Apollo creed coming out retirement in rocket four.

A little, a lot of shape.

get a lot .

of shape. A they could be a little lot of shape .

and also may not look like ivan drago. But but this is one shot character. This is one shoot character. I mean.

all he's fit. He's been in the arena.

Yeah he he's a multi time founder who at the top of Y C, and got to see everything that worked, yes, and got to see on the research. And he's been plugged away at this for what like the years. So there there's a big I didn't think there's a big gap to catch up on.

Now google has all the resources in the world, and they've got a appropriator assets to, and y've got all the intent of in the world. So do I think that google will be one of the top four players in A? I absolutely. But this idea there is going to come in steam roll OpenAI am.

have a production. I got a prediction within next year, Larry and sergey take the title of co ceos, and then they do a demo day where the two of them get on stage, and they actually do the demo process, if that happens. Fictional dic that's IT that's listen and base us to for present. Can you imagine of Larry freeman where the chances of Larry and sergey taking coc or lots of prediction one and then prediction to where are the chance of them running the next google I O, where they get on stage and they walk people through all the products that they shepherded and that they have a vested interested in that they want a demo.

There is an institutional problem at google at the top level, which does need to be solved, which is this position of constantly being in defense against the scrutiny, again, of regulators in public policy folks in, you know, all these different groups that are against google. And so as a result that the kind of cultural seasoning, particularly the executive and the board level, has been one of like, you know, protect the nest, don't overreach, don't overstep and it's a real you know I think one for the for the business school books or whatever, uh ultimately is what they end up doing about IT because now is a you know the time when that defensive posture is really of putting up the entire business at risk.

The same of microsoft member in the late nineties, when they got crush by that anti trust lawsuit, they made theory defensive.

Well, that cannot, but that can. Centrally, they put, they had a wartime sea come in. Bomber came in, and you followed by kind of an innovative guy who could can continue to build.

And I think that there may be a moment here I look, I love s and our, he is a great guy. great. You and I told you this, and I stated google the same day we were both in the same new glass class we were the frequent had on the T G I F day stage.

He was a product manager. Now he runs the company. I think the question is like hot, whether IT the C E O or the broader whole kind of executive org or the board, a degree of disruption necessary to shift their cultural seasoning is so necessary right now for them to have a shot of this. And similar what you just said, sex, like you're gonna need a bomber type moment to kind of reinvigorate this.

But it's important .

point when bomber trucked over during that period, after gates, when they were on their heals, he basically just focused on revenue and paying dividends and stock by backs. And the stock when I was, and he missed mobile.

And now if you forgetting one big thing, which is that was also because he adopted under concentrate to the D O. J exact, so the product managers of microsoft were replaced with lawers from the department, and you had to get their sign off before you could chip anything. So we have to remember that those things probably slowed microsoft down as well.

And the great thing that thought your head was a blank slate and the removal of that concentric, so he was able to do everything that just made a lot of sense and he's executed policy. I think the problem at google is not tonda or Larry, your surgeon. I think it's more in the deep bowls of medal management of that company, which is that there's just far too many people that probably have an opinion and their opinion is not shrouded in survival.

Their opinion is shrouded in elite language around what is the moral and ethical implications of this, and where has this been properly tested on the dia osburgh of nineteen different ethnic tribes of the amazon. That's the kind of decision making that is a nice to have when you are the second, third most valuable technology company in the world. But you have to be able to pause that kind of thinking and instead get into wartime survival mode, and it's very hard.

So IT doesn't almost matter at this point what sooner wants. The real question is what is the capability of middle management to either do IT or get out of the way? And I think that in all of these big companies that struggle, what you really see is an inability for middle management to get out of the way.

Or Franklin, just you need somebody to then fire them. And if you look at folks who get their group back to see what facebook does, what are they targeting? They are targeting metal management. If you look at what elon does in the companies at oes, there is virtually no middle management. But I get out of the way.

build product.

build product and ship IT yeah and what is the core truth? And so if failure is there in front of you, and if David is right, that you have two hundred million users come nowhere who are voting everyday with their time and attention to use an APP and that doesn't create a fireman fire where you get little management out of the way, and you are the senior, most people talking to the people doing the world and shipping things every day. You you are toast.

You are toast. A lot of people are starting to think we're moving a little bit too fast when IT comes to open A S incredible performance, which are GPT for the plugins on all this.

And so the future of life institute, which was formed in twenty fifteen, is a nonprofit that's focused on the risking major technology like A I and they did a petition titled pause, giant AI experiments and open letter, a bunch of computer scientists, a scientist letter and a the letter quote says, we must ask ourselves, should we let machines flood our information channels with propaganda and untruth? Should we automate away all the jobs, including the fulfilling ones? Should we developed on human minds that might eventually outnumber, outsmart, absolutely and replace us? Should we risk loss of control of our civilization? A number of notable tech leaders like elon sive wasna and a handful of deep mind researchers have signed IT.

What do you guys think of the letter? Are we going to slow down or not? Then we could ask the question generally, how closer we getting to agi, which is whatever be scared is that these agents start working with each other in the background to do things that are against human interest. I know that sounds like science fiction, but there is a theory that when these ais start Operating on their own, like we explained in the previous sort of segment here, up with plugging and they make agents that are Operating based on feedback from each other, could they get out of control and be most serious and then work against human interest? So what do you think, sex?

I think there's a difference between what could happen in the short term and then what could happen in the long term. I think in the short term, everything we're seeing right now is very possible. And let me to give you an example.

There was a really interesting tweet storm about a guy who wrote about how ChatGPT saved his dog. Did you guys see this? This is one of the really mind blowing ones to make use cases. So his dog was sick. Took him to a vet, vet prescribed some medication three days later, dog still seek, in fact even worse.

So the owner of the the pet was early copy and paste the lab result for the blood test for the dog with all the the lab values into ChatGPT and said, what could this be like? What's your likely diagnosis? ChatGPT gave three possible answers, three illnesses.

The first one was with the vet basically a diagnosed with, so that wasn't IT. The second one was excluded by another test. So he then went to a second vet and said, lest I think my dog has the third one and vet pressure ribes something, and sure enough, dog is current saved.

So that's really mind lowing that even though ChatGPT has specifically optimized, as far as we know for lab results, IT could figure this out. The reason I matching this is that gives you a sense of the potential here to cure disease to, you know, like I could see major medical breakthrough based on the A I. In the next five or ten years.

Now the question is like what happens in the long term, you know, as the AI gets smarter and smarter and we are kind of getting into the real science fiction, but here would be, the scenario is you're on ChatGPT ten or twenty or whatever IT is, or maybe some other companies, ai. And the developers asked the A I, how could you make yourself Better? Now do IT, which is a question we ask ChatGPT all the time in different context.

And so ChatGPT will already have the ability write perfectly by that point. I think, you know, code writing is one of the, I think, of its superpowers already. So IT gives itself the ability to rewrite this code, to auto update IT, to recursively make itself Better.

I mean, at that point, isn't that like a speciation event doesn't have a very quickly lead to the singularity? If the AI has the capability to rewrite its own code to make itself Better, and you won't IT very quickly write billions of versions of itself, and you is very hard to predict what that future looks like. Now I also don't know how far away we are from that. That could be ten years, twenty years, thirty years, whatever. But I I think the question worth asking .

for sure is IT worth slowing down the sex. Should we be pausing? Because based on what you said, said you, I think you've framed IT properly.

When these things hit a certain point, and they start reinforcing their own learning with each other, they can go at infinite speed, right? This is not comparable to human speed. They could be firing off millions, billions of different, I think.

scenario were definitely now on this. Fuck around. Find out. yeah. And so there is only one way to really find out, which is somebody's gone to push the boundaries.

The additive dynamics will get the Better of some start up. They'll do something that people will look back on and say, well, that was a little that was a bridge too far. So yeah, but it's just a matter of time. Yeah, I think we're not .

going to slow down. I actually think it's going the other way. I think things gonna speed up. And and the reason they're going to speed up is because the one thing silicon valley is really good at is taking advantage of platform shift.

And so when you think about like all the VC and all the founders, you know everyone accuses us of being leming. And so when there is like kind of like a fake platform for people kind of glam on something that ends up not being real, everyone is kind of got egg on their faces. But the flip side of that is that when the platform shift is real, silk valley is really good at throwing money at IT.

The talent knows how to go after IT, and they keep making IT Better and Better. And so that's the dynamic rain. Right now.

You look at seventy percent of the last yc class was ready. All A I startups. sure. The next one I be not to five percent.

So I think that we're on a path here where th Epace o f i nnovations a ctually i n t he s peed u p c ompanies a re g oing t o c ompete w ith e ach o ther. They're going to seek to invent new capabilities. And I think that the results are going to all be incredibly positive for some period of time. Like, you know, the vent example, we're going to cure illnesses. We're going to solve major .

problems positive than we invest more, we trust more. But the paradox of that is true. Thus pointing our free berg is if we trust IT more, we invest more.

And some person in a free market is going to say, you know what I need to be chat. P. T. Therefore, i'm going to take the rails of this thing. I'm going to let IT go faster and take off some constraints because I need to win, and i'm so far behind. How do you feel about that scenario, that sort of tremazan sexed up freeway?

I think there's like GPT three I think ran on seven hundred gigs. Is that right? Does anyone know GPT for run on? It's gotto be on some number.

It's you know not to not not a many multiple of that. but. Look, someone could make a copy of this thing and forget and develop an entirely new model. I think that's what's incredible about software and digital technology.

And also kind of you know IT means that it's very hard to contain similar like what we've seen in in biology ever since pyo logy got digitize through D N A sequencing and the ability to kind of express molecules through the editing. You know you can't control or contain the ability to do gene editing work at all because everyone knows the code. Everyone can make crisper cast molecules.

Everyone can make genetic systems in any lab, anywhere. Once I was out, I was out. And there's hundreds of variants for for doing gene editing, many of which are much improved over a crisp cast nine. I use that as an analogy because IT was this break through technology that allowed us to precisely specifically edit genomes and that allowed us to engineering biology and do these incredible things where biology effectively became software. And remember, crisp or cast nine gave us effectively, uh uh word processing type to will find in replace i'm the tooling that evolved from that is, is much Better.

So whatever is underlying, whatever the parameters are for GPT, for whatever that model is, if a close enough replicant of that model exists, a copy of that model is made and the new training data, new evolution, can be done separately. You could see many, many variants kind of emerge from here. And I think this is a good going of Timothy point.

We don't know what's ultimately gona win is there are enough of a network effect of a plugged model, as x pointed out, to really give OpenAI the sustaining competitive advantage. I'm not sure the model runs on seven hundred gigs that's less data than, you know, fits on my iphone. So you know, I could take that model, I could take the premiers that model and I could create an entirely new version.

I could for IT, and I could do something entirely you with that. So I I don't think you can contain IT. I don't think that this idea that we can put in place some regulatory constrains and say it's illegal to do this or you know try you know create I P around IT or protections around IT is realistic at this state. The power of the tool is so extraordinary. The extendability of the tools is so extraordinary.

So the economic and the of the various incentives are there for, you know, other models to emerge and whether they're directly copied from someone hacking into open a eye servers and making a copy that model, or whether you know, open source, or whether that someone generates something that ninety five percent is good and then its forks in a whole new class of models emerge. I think this is like it's a sax pointed out, highlighting the kind of economic market approving social Operating potential. And many models will will, will start to kind of come to to market.

What do we think the impact of White collar jobs getting anio latest by this technology.

if that in fact I want to do?

I just get one example here. So here's a red post that I was made aware of earlier this week. I lost everything that maybe love my job through mid journey overnight. I am employed as a radio artists in a small games company of ten people.

Our team is two people who based explains, he says, since mid journey version five came out, he's not an artist anymore, nor a 3d artist。 All they do is prompting photoshopping and implementing good looking pictures. And he basically says this happened overnight and he had no choice. Boss also had no choice.

He says, I am now able to create rig and animated character that spit out from mj a mid journey in two, three days before he took her several weeks in three d the differences that he cares about his, you know, job and for his boss is just a huge time money saver. He's no longer making art. And the person who was number two in the organization who didn't make a good content to him, is now embracing this technology because IT curries favor with this.

Ann's basically saying, getting a job in the game industry is already hard, but leaving a company in a nice team because A I took my job feels very good. Sop an I doubt I would be Better in a different company. Also, I am between grief and anger, and I am sorry for you. Art felt yet another .

reason fig ma really needs to close this acquisition from adobe. I mean, let's like the value of these apps are just getting gutted. If you take a workflow management tool for things like design and imaging and you reduced by an order of ninety percent, it's like, what is that APP experience worth? And how could you replicate that if you were a big company that already has distribution? That's one comment.

But what I would tell you, Jason, to answer the White collar question is I think there are a handful of companies you need to look at exclusively because they will be the first ones to really figure out how to display human labour. And that is T C S, sota consulting services, accenture, cognizant. These are all the folks that do coding for higher work at scale.

I think a center has something like seven hundred and fifty thousand employees. So the incentive to sort of freeze all packs to create Better utilize sation rates to increase profitability, it's quite obvious that always has been. They will be the first people to figure out how to use these tools at scale before the law firms of the accounting firms or any of those folks even sort of try to figure out how to display White color labor. I think it's gonna the coding jobs and it's going to be the coding for higher jobs that companies like accenture and T.

C. S. So those business processes do for other people develop kind of folks. They're onna need half as many people, twenty five percent as many people. We're going to find out the efficient front.

You yeah I see a different way. I mean, this argument that productivity least the job loss has been made for hundreds of years and it's always been refuted. When you make human beings more productive, IT leads to more prosperity.

More wealth grows, more growth. And so yeah, it's easy to think about in a narrow way, the jobs are going to be displaced. But but why would that be? Is because you're giving leverage to other human beings to get more done.

And some of those human beings, really, anybody with a good idea, is not going to be able to create a start a much more easily. So you're gna see a huge explosion in creativity, in start creation, new companies, new jobs. Imagine think about the case of, you know, zark board founding facebook and harvard.

He wrote the first version himself, maybe with a couple of friends. That project happened and turned into a giant company because he was able to self execute his idea without need to raise venture capital or even recruit employees, even really before. For me, a company, anyone with a good ideas, need to do that. Soon you're going be able to use these A I tools. They truly will be no code you'll build to create an APP or website is like speaking to some AI program, natural language .

way more flowers will bloom.

more create IT will create, I think, a lot of this location. But for every testimonial, that is like the one that you showed, which I think is, i'd say, a little bit overly dramatic. I have seen ten or a hundred testimonials from quoters on twitter or other blogs talking about the power that these new tools give them.

They are like, this makes me a tennis engineer, right? And especially these, like junior engineers, are right at school who don't have twenty years of coding history. They get superpowers right away that IT .

makes them so much Better. Let me give .

you response to that kind. So so and using sex as point, that guy saying, what used to take me weeks I can now do in two to three days, and I feel like my work is gone. And that's because he's thinking in terms of his output being static.

And if he thinks about his output being dynamic, he can now, in the matter of three weeks, instead of making one character, he cannot make a character every two days. So he can make thirty characters in three weeks. That's an alternative way for him to think about what this tooling does for him in his business.

The number of video games will go up by ten ex or a hundred x or a thousand x. The number of movies and videos s that can be rendered in computers can go up by ten x or hundred x or thousand x. This is why I I really believe strongly that at some period of time, we will all have our own movie or our own video game ultimately generated for us on the fly based on our particular interest.

There will certainly be shared culture, shared teams, you know, shared morality, shared things that that, that tie all these things together, and that will become the shared experience. But in terms of like us all consuming the same content, IT will really like you with youtube and tiktok or all consuming different stuff all the time. And this will enable an acceleration of that evolution and personalization.

I'll also highlight, you know, back in the day, one human had to farm a farm by hand, and we actually got the tool of a hole, and we can built on the ground and make, you know, make them faster. And we ve got a law, and then we got a tractor. And today, agricultural farm equipment allows one farmer to farmer over ten thousand accord. You go to western australia, it's incredibly, guys have four world planters and harvesters and it's completely change the game. So the unit of output performer is now literally millions of times what IT was.

just a hundred. And in that case, nobody wants to do back breaking labor in the fields, and everybody wants that. But in this case, let me just read one quote that I didn't read in the original reading of this.

He says I want to make art that isn't the result of script t internet content from artists that we're not as. And so I think that's part of this is that it's a spoke art. But I the one question I have for sex, sex you we started this conversation, we saying, hey, this is different than anything in terms of efficiency that came before.

This is i'm going to a put the words mattheo, but this is like a step function, more efficient. So to the argument of, hey, efficiency has always resulted in, you know, more ideas and and we found something to do with people's time. Is this time different potentially? Because this is so much more powerful? This isn't just like a spell checker.

I would say differently, I think. And I agree with what jack l sing, because I think that the thing that technology has never done is try to display human. It's allowed us to replace physical exertion of energy.

But IT is always preserved. Humans injecting our judges. And I think this is the first time where we are being chAllenged with autonomic systems that has some level of judgment.

Now we can say, and it's true, can reform months in that, that judgment isn't so great. But eventually, and because of th Epace o f i nnovation, eventually, probably not that far away, the judgment will become perfect. I'll give you totally different example.

You know, how many pilots are there in the world? Will we at some point in the next ten years, want folks to actually manually take off and land? Or will we want precision guided instrumentation and computers and sensors that can guarantee a pitch perfect landing every single time in all kinds of weather conditions, so that now plans can even have fifty x the number of sensors with a computer that can then process IT and act accordingly? Just a random example that isn't even thought of when we talk about sort of where A I is gonna AR its head.

I think that this judgment idea is an important one to figure out, because this is the first time i've seen something that is bumping up against our ability to have judged. And what this person was talking about in this mid journey example is his judgment has been using ed. yes.

Yeah, I would disagree. So yeah, let me just let me just make one point on this. So you know, an image is a matrix of, you know, data that's rendered on a screen and as pixel and those pixels are different colors. And you know that's what an image is.

Is IT or is IT? Is IT the judgment of the creator?

Well, no, i'm just an image in general, so don't be photoshop in digital photophoretic. SE photographers were like, this is, you know, B S. Why digitalized photography with analog beautiful before? And then what digital photography allowed is the photographer to do editing and to do work that was creative beyond what was possible with just a natural photograph taken to a camera.

And there are arguing different art forms, but IT was a new kind of art form that emerged through digital photography. And then in the usually ninety, there was a plugin sweet called kite power tools that came out in a dob, photoshop. And IT was a third party pluggin and said, you would, you would buy IT.

And then I would work on photoshop. And I did things like motion blur, sharpening pixels, lation. All these interesting kind of like features.

And prior to those tools coming out, the judgment of the digital artist, the digital token pa was to go and do pixel by pixel changes on the image, to make that pixel, to make that image look blurry, or to make IT look sharper, or to make IT look like IT had some really interesting motion feature. And the sky's power tools created this instant tooker were in a few seconds you created a blur on the image. And that was an incredible tool kit.

But a lot of digital artists said, this is automating my work. What is my point now? Why am I here? And the same happen in the animation when three um when you know C G I came around and animators were no longer animating cells by hand.

And in every point in this evolution there was a feeling of loss initially but then the evolution of a whole new art form emerges and an evolution of a whole new area of human creative expression emerged. And I think we don't yet know what that's gna look like. Do you think you .

think the the level of judgment that A I offers you is the same as the level of judgment that kyp wer tools offered to? Yeah look.

I mean, I think that the person making the judgment of the decision which pixel to change into what color felt like, you know, I have control .

and I think it's ultimately sagan with I think I think that a mag different .

is more a give .

you a different example. Today, we use radiologists and pathologists to identify cancer. You there are close loop systems we have won right now. That's in front of the F, D, A. That is a total close loop system that will not need any human input.

So I don't know what those folks do, except what I can tell you is that we can get cancer detection basically down to a zero percent rate that is not possible with human intervention. That is judgment, right? So I just think it's important to really acknowledge that this is happening at a level that is never happened before.

You may be right that there's some amazing job for that radiologist or pathologist to do in the future. I don't know, i've hand what that is, but these are closed loops systems. Now that think for themselves and self improve.

I get IT. But I think that there there is an unsafe set of things that emerge. We did not have the concept of instagram influencers. We did not have the concept of personal trainers. We did not have the concept of like all these new jobs that have emerged in the past couple of decades that people enjoy doing that. They can make money doing that. There is a greater kind of experience and level of fulfilment for those that choose and have the freedom to do IT then what they were having to do before when they had to work just to .

make money and good that radiologist or pathologists want to do, be a trainer or .

polite instructor. No, I think don't know AI having the same.

You you have any thoughts on this as we rap this topic? It's obviously a lot of passion coming out. Yeah, the elimination of White color jobs in a massive way.

I think that this a short term versus long term thing. In the short term, I see the benefits of AI being very positive because I don't think it's, in most cases, wiping out human jobs. This is making them way more productive.

You still need the developer is that there are five times or ten x more productive. But I don't think we're at the point in the short term, we're going to eliminate that role entirely. And what i've seen in basically every start of being a part of is that the limiting factor on progress is always engineering.

Being with that is always the the thing that you wish you had more of. Its the product road map is always the most competed on thing inside the organization. Everyones try to you know get their project, prioritize because there's this never enough. And during being with it's really the lifeblood of the company.

So if you make the developers more productive and maybe you just accelerate the point road map, I I don't think in the short term that was going to happen is these companies you're going to look to cut all their developers because one or two of them can do ten times to the work. I think that they're gona try and accelerate their product road maps. Now again, you have this long term concern that maybe developers at all at some point, but I think that the benefits of developing this technology are so great in the short to mid term that we're going down that past no matter what. And we're just going to find out what that long term really looks like and maybe the long term will look very different. I mean, once when you get past the short term, we may have a different long term view.

I think in this narrow vertical, I A hundred percent agree with you. Look, I I think that A I is gone to eliminate unit testing. IT is already done.

So it's going to eliminate most forms of coding. The engineers that you have, all of them will now become ten x engineers. So with fewer of them or with the same number, you'll be able to do as much or more than you could have before.

That's a wonderful thing. And all i'm saying on that specific narrow vertical is you'll see the first real tet and companies like accenture and tcs because and cognition because they have an an immediate incentive to use this tooling to drive efficiency and profitability that's rewarded by shareholders. It'll be less visible in other company.

So but what I am saying that is that you have to think about the impact on the end markets for a second. And I think that A I does something that other technology layers have never done before, which is supply human judgment in a close loop manner. And I just think it's worth appreciating that there are many systems and many jobs that reply, that rely on human judgment, where we deal with err bars and an error rate, that a computer will just destroy and blow out of the water.

And we will have to ask ourselves, should this classic job exist with its inherent error rate? Or should I get replaced link by a computer which has no arrow? And I think that's an important question that we're putting on the table.

okay. So let's rap here. I just have my final dot unit is like you're going to see entire jobs categories of jobs go away.

We've seen this before, phone Operators, travel agents, copy editors, illustrators, lobby designers, account and sales development reps. I'm seeing a lot of these job functions in the modern world like phone Operators previously. I think this could wholesale just go away and they would just be done by A I.

And I think it's going to happen in a very superior time. So gonna about who can transition and some people might not be able to make the transition. And that's going to be pain, suffering, and it's going to be in the White color rides and those people have more influences. So I think this is could lead to some social disturbance with few sex.

I'm going to learn politics and be an influencer that's IT.

But I do agree with sex that the software development backlog, this is what you're saying, is so great that I don't think we'll see in software development for a decade or two. There's just so much software that still need to be made. At last week, talk about tiktok and the first bipartisan hearing, you've seen a long time and people actually, I think, framing correctly, exactly how dangerous he is in my anuath tiktok in the united states.

And of course, then we get the great disappointment of the actual bill they restrict act was proposed by a senator, mark Warner, democrats, Virginia, on march seventh. The problem with that is as that seems like it's poorly worded, that there will be civil penalties and criminal penalties to americans for breaking the law in using software that's been banned. And many people said, you know, this probably is just bad language I have a question.

yeah, does does this apply to incognito mode? Kind of night?

Yes, yes.

they're saying, they're saying that you can get know you can get fined or twenty years in jail whatever IT is for using a VPN to VPN to take talk facebook. What are your thoughts on IT?

Look, I think this is a real threat to the open internet. I'm really concerned about the language that's been used that basically speaks to protecting the safety and security of american people by actively monitoring network traffic and making decisions about what network traffic isn't isn't allowed to be transmitted across the open internet.

It's the first time that I think in the united states we are seeing like a real threats and real set of having or from our government that looks and feels a lot like what goes on in china and elsewhere where they Operate with a close internet. And internet is a controlled, monitored, observed tract and and gates are decided by some set of administrators. And what isn't isn't appropriate.

And the language is always the same. It's for safety and security of the people. The in entire purpose of the internet is that I did not have bounds, that I did not have governments, that I did not have controls, that I did not have systems that are politically and economically influence of p architecture.

The internet was and always would be open. The protocols are open. The transmission of data on that network would be open, and as a result, all people around the world would have access to information of their choosing, and IT allowed ultimate freedom of choice.

You know this this kind of is the first of what i'm concerned, creates a precedent that ultimately leads to a very slippery slope saying that tiktok cannot make money in the U. S. By charging advertisers or managing commerce flows is one thing that's where the government can and child, and could, if they chose to, you have a role. But I think going in and and observing, tracking internet traffic and making decisions about what is isn't appropriate for people, I think is one of the things that we all should be most concerned about what's going on right now.

There is no end in sight to this ah if you allow this to happen uh in the first time, you know V P S uh virtual private networks allow you to anonymous ly access internet traffic and and an access internet traffic via remote destinations so so that the ultimate consumption of content that you're using can be track and monitored by local agencies or I S P. And I think that saying that that can now be restricted takes away all ability to have true privacy and all rights, uh, to privacy on the open internet. So i'd love to talk about this more unfortunate.

I got to run. This is a super threat to me. And I, I I think this is something we should be super, super concerned about and that the entire community of technology, internet and anyone that wants to have your freedom of choice steps up and says this is totally uh, inappropriate and over which there are other ways to manage .

stuff like complete over age .

sex ah the intentional .

over each are poorly written or summary between both .

I think both I think this is the biggest bitten switch that washing the central government ever tried to pull on us. Everybody thinks that they are just trying to ban tiktok from Operating the U. S if that's all they did, then I think the bill would be supported by most americans.

But that's not what they're doing. They're not restricting tiktok. They are .

restricting us. That's not the goal here. Yeah but a bay switch.

It's a huge brain switch. And so just you know what the approvals is that a us. Citizen using a VPN to access tiktok could theoretically be subjected to a maxim penalty of one million in fines or twenty years in prison or both.

Now you'll say, you know mark Warner, the sponsored tional sare up and down down. That's not the intent, but the problem is that the language of the bill is so vague that some clever prosecutor may want to pursue this theory one day and that is be stopped. Also, there's another problem with the bill, which is you think this is just about tiktok.

It's not what they do is IT says here, I guess they don't wanted mention tiktok by names, so they trying to create a category of threatening application. But because this is a category, it's very, very broad. So the bill states that IT covers any transaction, transaction, not just an APP in which an energy described in super graph b has any interest, and then ended scribing.

Some paragraph b are called a foreign adversary and energy subjects, the jurisdiction of organising the laws of a foreign adversary and energy own director control by either of these. And then IT gives the executive branch the power to name a foreign adversary, any foreign government regime that one of the cabinet secretary defines without any vote of congress. So this is giving sweeping powers to the executive branch to declare, you know, foreign parties to be enemy .

like the plot of the prequel, great powers.

Here we go. Know, we criticize china having a great firewall. What do you think this is?

Yeah, I mean, this should obviously have nothing to do with the american consumer and everything to do with the foreign adversary collecting data of americans at scale. This could be written in a much .

simpler and that should be one sense, which is that APP stores are prohibited from allowing tiktok be an APP in their store. That's what they do in india. That case.

clothes game over OK, right? They they block like a hundred chinese apps, and I think they're society, still functioning. So, you know, all do respect to A O C. You know, like the idea that one hundred fifty americans, million americans, are gona suffer because they can be tracked by the ccp is kind of nuts.

This is gonna give sweeping powers to the security state to survey us, to prosecute us, to limit internet usage. This is basically the biggest power grab and beating switch they've ever tried to pull on us. And again, if they really were concerned .

about tiktok, it's one sense. Yeah, we were done. Aren't everybody spit an amazing episode for the time of science of berg, the dictator, chaos poy hop? Tia, I am the world's greatest moderator, and we'll see you next time. Byebye .

your.

Winter, man.

And we open sources to the fans and they .

just got crazy with.

Should all get a room, just have one big, huge or because like sexual attention.

我 一定 to get。