The Collapse No One Is Prepared For | Peter Thiel AI Masterclass

Well, I think you're probably not supposed to say. No, no, no. If you... I think, look, I think there's so many ways our policies are messed up, but probably the one thing that's roughly correct on the Taiwan policy is that we don't tell China what we're going to do. And what we tell them is we don't know what we'll do and we'll figure it out when you do it, which probably has the virtue of being correct. And then I think if you had a red line at Quemoy Matsu, the island five miles off the coast of China, that's unbelievable. If you say we want some guardrails and we won't defend Taiwan, then they get invaded right away. So I think the policy of not saying what our policy is, and maybe not even having a policy in some ways, is relatively the best. I think anything precise you say is going to just lead to war right away.

It's not worth World War III. And I still think it's quite catastrophic if it gets taken over by the communists.

Probably it's some very messy in-between thing, sort of like what you have in the Ukraine. What I think happens economically is very straightforward. I think basically you have with Russia and Germany you had one Nord Stream pipeline, and we have the equivalent of a hundred pipelines between the US and China and they all blow up. You know, I met the TikTok CEO about a year ago and, you know, maybe I wouldn't have said this now, but what I told him and what I felt was very honest advice was, you know, you don't need to worry about the US. We're never going to do anything about TikTok. We're too incompetent. But if I were in your place, I would still get the business out of China. I would get the computers out, the people out, I'd completely decouple it from ByteDance because TikTok will be banned 24 hours after the Taiwan invasion. And if you think there's a 50/50 chance this happens, that will destroy 100% of the value of the TikTok franchise.

You know, he said that they had done a lot of simulations and there were a bunch of companies in World War I and World War II that managed to sell things to both sides.

He didn't disagree with my frame, and so I always find that flattering if someone basically agrees with my framing. So he seemed perfectly bright to me, even though... not game theory, but a lot of bright people.

Well, I think look, there are a lot of different ways in which businesses are coupled to China. There were investors that tried investing, there are people who tried to compete within China, there are people who built factories in China for export. And you know, there are different parts of that that worked to varying degrees. But yeah, I certainly would not try to invest in a company that competed domestically inside China. I think that's virtually impossible. I think it's probably quite tricky even to invest in Chinese businesses. And then there is this model of building factories in China for export to the West, and it was a very big arbitrage. These things do work. You know, I visited the Foxconn factory nine years ago and it's, you know, you have people get paid a dollar and a half, two dollars an hour and they work 12 hours a day and they live in a dorm room with two bunk beds where you get eight people in the dorm room, someone's sleeping in your bed while you're working and vice versa. And you sort of realize they're really far behind us or they're really far ahead of us. And either way, it's not that straightforward to just shift the iPhone factories to the United States. So I sort of understand why a lot of businesses ended up there and why this is the arrangement that we have. But yeah, my intuition for what is going to happen, without making any normative judgments at all, is it is going to decouple.

It's presumably pretty inflationary. Yeah, that's probably the... you'd have to sort of look at what the inelasticities of all these goods are.

It's probably not that... it may not be as inflationary as people think because people always model trade in terms of pair-wise, in terms of two countries. So if you literally have to move the people back to the US, that's insanely expensive. I don't know how much it would cost people to build an iPhone, but you just... well, I think India is sort of too messed up, but you shift it to like Vietnam, Mexico, there are five billion people living in countries where the incomes are lower than China. And so probably the negative-sum trade policy we should have with China is we should just shift it to other countries, which is a little bit bad for the US, extremely bad for China, and let's say really good for Vietnam. That's kind of the negative-sum policy that's going to manifest as this decoupling happens.

Well, it's not just up to us. So, yeah, there's... and so I don't know. It's obviously somewhat of a black box. We don't exactly... I feel we just have no clue what people in China think. But I think it's sort of the sense of history is strongly the Thucydides Trap idea that you have a rising power against an existing power and it tends to... you know, it's Germany versus Britain before World War I and it's Athens against Sparta, the rising power against the existing power, you tend to get conflict. That's probably what deep down I think is really far in the China DNA. So I'd say maybe the first... I don't know, the meta version would be the first step to avoiding the conflict would be we have to start by admitting that China believes the conflict's happening, right? And then if people like you are constantly saying, well, we just need to have some happy talk, right? That is a recipe for World War III.

Obviously, in general, you know, I don't know. I'm not sure Trump should have talked to the North Korean dictator, but yeah, in general, it's probably a good idea to try to talk to people even if they're really bad people most of the time. And you know, it's certainly a very odd dynamic with Russia at this point where I think it is impossible for anybody in the Biden administration even to have a back-channel communication with people like... I don't think Tucker Carlson counts as an emissary from the Biden administration. And if anybody gets Tuckered, or I don't know what the verb is, who talks, you know, that seems worse than the alternative.

Oh man, that's sort of a big question. I always had this riff where I don't like the buzzwords and, you know, machine learning, big data, cloud computing, you know, I'm going to build a mobile app bringing the cloud to... if you have sort of a concatenation of buzzwords, my first instinct is just to run away as fast as possible. Some really bad groupthink. And for many years, my bias is probably that AI was one of the worst of all these buzzwords. It meant the next generation of computers, the last generation of computers, anything in between. So it's meant all these very different things. If we roll the clock back to the 2010s, probably the AI, to the extent you concretize it, I would say the AI debate was maybe framed by two canonical books. There was the Bostrom book, Superintelligence 2014, where AI was going to be this superhuman, super-duper intelligent thing. And then the anti-Bostrom book was the Kai-Fu Lee 2018 AI Superpowers, you can think of the CCP rebuttal to Bostrom, where basically AI was going to be surveillance tech, face recognition, and China was going to win because they had no qualms about applying this technology. And then if we now think about what actually happened, let's say with the LLM and ChatGPT, it was really neither of those two. And it was this in-between thing, which was actually what people would have defined AI as for the previous 60 or 70 years, which is passing the Turing test, which is the somewhat fuzzy line. It's a computer that can pretend to be a human, or that can fool you into thinking it's a human. And even with the fuzziness of that line, you could say that pre-ChatGPT wasn't passed and then ChatGPT passed it. And that seems very, very significant. And then obviously it leads to all these questions. What does it mean? Is it going to complement people? Is it going to substitute for people? What does it do to the labor market? Do you get paid more, paid less? So there are all these questions, but it seems extremely important. And it's probably, you know, certainly the big picture questions, which I think Silicon Valley is always very bad at talking about, is like, what does it mean to be a human being, right? The stupid 2022 answer would be that humans differ from all the other animals because we're good at languages. If you're a three-year-old or an 80-year-old, you speak, you communicate, we tell each other stories. This is what makes us different. And so yeah, I think there's something about it that's incredibly important and very disorienting.

It's pretty confusing and I think, I don't know, this is always where I'm anchored on the late 90s as sort of the formative period for me. That I keep thinking that AI in 2023, 2024 is like the internet in 1999. It's really big. It's going to be very important. It's going to transform the world not in six months but in 20 years. And then there are probably all kinds of incredibly catastrophic approximations where, you know, what businesses are going to make money, who's going to have the monopoly, who's going to have pricing power is super unclear. Probably one layer deeper of analysis, if attention is all you need and if you're not post-economic, you need to pay attention to who's making money. And in AI, it's basically one company is making... Nvidia is making over 100% of the profits, everybody else is collectively losing money. And so there's sort of a... you should try to do some sort of analysis. Do you go long Nvidia? Do you go short? Is it a really durable monopoly? And then it's hard for me to know because I'm in Silicon Valley and we haven't done anything in semiconductors for a long time, so I have no clue.

I don't know. I don't know if this is the same, but this is the history of 250 years, the industrial revolution. What was that? It adds to GDP, it frees people up to do more productive things. You know, maybe there was... I know there was a Luddite critique in the 19th century of the factories that people were going to be unemployed and wouldn't have anything to do because the machines would replace the people. Maybe the Luddites are right this time around. I'm probably pretty skeptical of it, but yeah, it's extremely confusing where the gains and losses are. There probably are... there's always sort of a hobby horse. You can always just use it on your hobby horses. So, I don't know. My anti-Hollywood or anti-university hobby horse is that it seems to me that the AI is quite good at the woke stuff. And so if you want to be a successful actor, you should be maybe a little bit racist or a little bit sexist or just really funny. And you won't have any risk of the AI replacing you. Everybody else will get replaced. And then probably, I don't know, Claudine Gay, the plagiarizing Harvard University president. The AI will produce endless amounts of these sort of... I don't even know what to call them, woke papers. And they were all already sort of plagiarizing one another because they were always saying the same thing over and over again. They were using their own version of... going to flood the zone with even more of that. And you know, I don't know, obviously they've been able to do it for a long time and no one's noticed. But I think at this point it doesn't seem promising from a competitive point of view, which is obviously my hobby horse, so that's maybe just wishful thinking on my part.

You know, I always think you want to instantiate it more in companies than things. If you ask where is innovation happening, in our society, and it doesn't have to be this way, but it's mostly in a certain subset of relatively small companies where you have these relatively small teams of people that are really pushing the envelope. And that's sort of what I find inspiring about venture capital. And then obviously you don't just want innovation, you also want it to translate into good businesses. But that's where it happens. It somehow doesn't happen in universities. It doesn't happen in government. You know, there was a time it did. I mean, somehow in this very different country that was the United States in the 1940s, you had somehow the army organized the scientists and got them to produce a nuclear bomb in Los Alamos in three and a half years. And the way the New York Times editorialized after that was, you know, it was sort of an anti-libertarian writeup. It was, you know, there were... obvious maybe if you'd left the prima donna scientists to their own, it would have taken them 50 years to build a bomb and the army could just tell them what to do and this will silence anybody who doesn't believe the government can do things. And they don't write editorials like that in the New York Times anymore. But I think that's sort of where one should look. I think a crazy amount of it still happens in the United States. And there sort of... we've episodically tried to do all this investing. We probably tried to do too much investing in Europe over the years. It's always sort of a junket, sort of a nice place to go on vacation as an investor. And it's very... I don't have a great explanation but it's a very strange thing that so much of it is still... the US is somehow still the country where people do new things.

Yeah. Well, this is always one of the big picture claims I have that we've been in an era of relative tech stagnation the last 40 or 50 years. The tagline, they promised us flying cars, all we got was 140 characters, which is not an anti-Twitter, anti-X commentary, even though the way I used to always qualify it was that at least it was a good company, you had 10,000 people who didn't have to do very much work and because you smoked marijuana, all very similar to Europe. And so I think that part actually did get corrected. But I think what went wrong, because you point out that it's not a technology trend tracker that you think about, it's about people and teams that innovate and drive to outcomes based on their view of the world. And what's gone wrong with our view of the world and our ability to organize to achieve the seemingly unachievable with very rare exceptions. Obviously Elon's here later, but yeah, it's overdetermined. The rough frame I always have, and again it's not that there's been no innovation. There's been a decent amount of innovation in the world of bits, computers, internet, mobile internet, crypto, AI. So there are sort of all these world of bits places where there was significant but sort of somehow narrow cone of progress. But it was everything having to do with atoms that was slow. And this was already the case when I was an undergraduate at Stanford in the late 80s. In retrospect, any applied engineering field was a bad idea. It was a bad idea to become a chemical engineer, a mechanical engineer. Aeroastro was terrible. Nuclear engineering, everyone knew, I mean, I think no one did that. And there's something about the world of atoms that, from a libertarian point of view, you'd say it got regulated to death. There probably... there's some set of arguments where the low-hanging fruit got picked and it got harder to find new things to do. Although I always think that was just a sort of baby boomer excuse for covering up for the failures of that generation. And then I think maybe a very big picture part of it was that at some point in the 20th century the idea took hold that not all forms of technological progress were simply good and simply for the better. And there's something about the two world wars and the development of nuclear weapons that gradually pushed people into this more risk-averse society. And it didn't happen overnight, but maybe a quarter century after the nuclear bomb. It's like by Woodstock it happened.

Yeah. Woodstock was three weeks after that.