Raoul Pal: AI is the Most Disinflationary Force in Our Lifetimes

0:00Last week on my other podcast, The Breakdown, I released a show with Real Vision CEO Raoul Paul. That show was about everything from global macro to Bitcoin and crypto to AI, where it fits in the broader economy, what it means for big companies, how it relates to other areas of disruption, and how we're going to come out the other side of it. The podcast that follows is the section of that conversation where we focus on AI. When the history books are written, this sort of six-month period, starting with the introduction of ChatGPT in November of last year, plus sort of the advance of Midjourney and Dali and Stable Diffusion, will be seen as the sort of inflection point moment.

0:44But particularly that five-week period that you referenced before, where it went from zero to 100 million users of ChatGPT, which was, you know, TikTok had previously been the fastest to 100 million users, and that happened over the course of eight or nine months. So, you know, 4x faster than that. You've obviously paid attention to AI before. It wasn't like it just hit you over the head. But what was your personal experience of saying or of recognizing that this was different than just another new technology that had popped up in another hype cycle? It was when an old friend of mine, Ahmad Mostak, who's now become quite well-known, Ahmad was a macro guy.

1:21I'd known him for years. He'd been on Real Vision, talk about emerging markets and stuff. Macro guy, very smart guy. And yeah, I'd known him for a decade or longer. And over the pandemic, he hit me up and said, hey, listen, I've got something interesting for you. I'm working with the WHO building pandemic AI modeling. And I think there's something interesting and we can model out the spread of the virus. So I'm like, great. So he came on and gave us a pretty grim assessment of where the virus was coming. This was February, March 2020. So he was really good. So anyway, thought nothing of it. And I saw him on Twitter starting to talk about, show some stable diffusion images and talk about AI.

2:04So I just hit him up. I said, what are you up to? So he said, look, we need to talk. So I got him on Real Vision, sat down and said, right, what the hell is going on? And he laid out what was happening in AI and stable diffusion and how fast this was happening and that he was launching and obviously ChatGPT 3 had launched. So he's like, well, this is kind of where it's going. Within three months of that hitting, which was a big shock to everybody, that interview was a huge shock. It kind of, everybody went, I did not know any of this. And then ChatGPT 4 came out and then it just went, you know, the biggest thing we've ever seen.

2:43So yeah, I was lucky just by random. I got into this, as you said, we've all been aware of this for a while but i had that acceleration moment with the first interview with em had where i'm like oh my god this is happening now and he's already telling us back then which was what november or something he was like yeah well text a video video to video he's like we can create fake characters he goes he's buying sports rights to some of the most famous people i can't really give it away of all times could trade ai so they can compete people who've never competed before. And it's like that, as you were talking about this warping, what is reality?

3:23Reality doesn't exist in the way that we understand it to exist anymore. And once you understand that, that was the first breakthrough for me. And then it was the understanding that this is the replacement of human knowledge at scale. So first we can augment ourselves, but it also replaces ourselves as well. And knowledge was not scalable, not in the same way. But this now becomes infinitely scalable, which is what people can't get their heads around yet. One of the things that makes this period that we've been living through so fascinating, and in particular, that moment, that zero to 100 million moment, was so often, even in crypto, especially crypto, crypto is a great example.

4:04These things that people who are deep in them understand as the future TM, right? Because of, you know, the inevitable trend lines, right? Like you laid out in the first hour of this conversation, all the reasons that, you know, that Bitcoin, that crypto, that Ethereum feel completely inevitable to people who are in them, right? It's so much bigger and more inevitable than any amount of Gensler screaming at Congress or vice versa can do, even if we have to live through that in the short term. It's why that, you know, when this interview is probably done, we'll probably have spent less time on quote unquote crypto than anything else, because it's just sort of like, it's just the inevitable consequence of this, right?

4:40That's the way the technology, new technology often feels for people inside it. But for people outside it, it's completely abstract, right? We still, the average person in America who has heard of Bitcoin and crypto and has no idea why it would be relevant for them, right? It's just sort of the way that it is. Now, that might be different in different kind of countries that have different sort of monetary regimes. But by and large, it's still a phenomenon where there's an early adopter set who clearly get it and everyone else. What was fascinating or what has been fascinating about AI, and in particular, again, like I said, the combination, I think, of chat GPT, a chat interface sitting on top of GPT-4, or GPT-3.5 even, and the ability to use the mid-journeys and DALIs and stable diffusions of the world has made people see way faster what the implications are for themselves.

5:34I mean, they feel it. Like if you go on TikTok right now and you watch the average video about AI, it is already, and we are less than six months after this happened, how AI, you know, how ChatGPT is going to change real estate forever because all of a sudden you have all these, you know, you don't need to write listings in the same way. And then it's how ChatGPT changes copywriting and marketing. And it is a absolute race for people to understand how their industry, their jobs, their specific skill set is going to change and be disrupted on the basis of this. And fascinatingly, it feels to me like the change is so profound to people that the people who are actually paying attention and who have dove in and even tried these things have almost entirely skipped over the step of being mad that it is likely to replace them and jumped right into how do I change?

6:24Because it feels so inevitable. Once you see it, it's impossible to unsee. There's no sort of like, no, we should stop and copywriters should form a union to stop copywriting AI, right? It's just, okay, how do I become the copywriter that's powered by AI? I guess, as you look at the AI space, how much do you think about the disruption in terms of either one, the transformation of work of individual kind of jobs and careers, So this kind of gets into the productivity piece. Two, the transformation of industries and larger sort of, you know, deflationary trends that might have been there before with technology, but are going even further.

7:02Or three, transformation of societies in terms of our, you know, understanding of purpose, in terms of structure of society, or in terms of things like safety and X-risk. You know, obviously, these are all kind of pieces of it. But where have you found yourself focusing, you know, as this is sort of blasted onto the scene? The easiest one to answer is the second one. This is the largest disinflationary shock the world will ever have. And it will keep playing out as these other exponential technologies become implemented. Just think, I was in New York when I was reading a thread on Twitter. I was in an Uber.

7:37And it was a thread about the rise of autonomous vehicles in California and how this guy, whether he's in the industry, he's around tech industries, like three years ago, I'd see a couple of these things. And now I see them every day, 10 times a day. And I looked around me in New York City, and every car was an Uber, a taxi, a bus driver. And you're like, oh, my God, AI, all of this stuff together is the biggest disinflation shock the world will ever face. The WTO agreement and China entering it was gigantic. But this is of a different order of magnitude. So that's good in some respects, but that disinflationary shock is all part of this work replacement.

8:18Why is it disinflationary? Because it replaces humans, and it replaces humans at a scale. So it's kind of like, I think why people are accepting of it so fast, because it's so inevitable, it's because there is no other choice. What are you going to do? Shake your fist at the sky? And we don't know what it means. We don't know what it means for work. We know that if you get it right, you augment yourself on a scale of which you cannot comprehend. Augmented humans is a very attractive thing. So we become super productive units. That's okay once the baby boomers have died off. It's not okay when we've got too many of us around still, because we just don't need as many humans.

9:01That's the hypothesis. Now, could it be that we create enough productive activity through this can change in productivity that we can employ more humans in these kind of areas? I don't know. But the problem is, we've got robots coming as well. So if the AI doesn't get your job, the robots will get your job. So it's a very, very scary place for work. But oh, my God, if you leverage this, I think it's a renaissance for both global economies and for people themselves. One of my hypotheses for Web3 was that the rise of all of this is leading to a need to support incomes for people. This is a universal basic income argument.

9:42There's ways of solving that, both the private sector and state sector. State sector is a problem because we don't have enough money, so we have to tax the robots maybe to get the money, but we're all underfunded. We've got this problem with debt. There is a way that you and I can restrict our data online because we have agency over our data, and therefore sell it to the platforms and the advertisers and get income for that. Okay, that's nice, but probably not enough. And I was talking to Yatsui, and I had a big breakthrough, which was universal basic equity, which is in Web3, in these decentralized digital nation-state societies, communities, you can own the assets or the currency of those communities, and you can be rewarded for being a good community member.

10:28And we're seeing that all over the place now. It's pretty nascent, but it's everywhere. And if that is the case, then we can replace our purpose. And our purpose could be society and community. Now, what that means, could it be for work output? Could it be for leisure output? It doesn't really matter. Could it be for cultural output? For sure. For sure, Louis Vuitton understand this. For sure, Nike understands this is the loyalty of Web3 and tokenized assets, whether it's a loyalty currency, community currency like Starbucks, or whether it's an NFT like Adidas, these things are making people participate in the success of those brands, those cultures.

11:15And we will see this in music, and we'll see it all over the place. So I'm actually interested in how do we adapt because we will not be able to all be supercomputing AI geniuses with amazing prompt skills. And even prompt skills go away pretty soon. And there'll be such a proliferation of newsletters. We're already seeing it. How many bloody AI newsletters are there? There's like 50. And they've all come and they're all being written by ChatGPT4. So there's excess content, excess of everything else. And we're going to need a purpose and a meaning and a place of trust. And a place of trust is these digital societies online.

11:49It's the digital network state idea of Balaji, something I've been talking about for a very long time, is this way that we live our digital lives. Like you and I, we've never met. But I've spoken to you so many times. I'll consider you a friend. We'll hear each other up. We'll chat about stuff. We will talk to each other in this digital representation. We live in the same community, which is somewhere the nexus of FinTwit and CryptoTwitter. and those are the overlaps. And we're both in the media area as well. So there's these number of overlaps of these digital societies that we both belong to.

12:27And that's very different a world than it was before 2008. That didn't exist. I mean, Facebook was 2012. So again, it's one of the answers for society is staring us in the face, which is we can find a purpose. And the purpose is humanity, culture, brand, businesses by doing it together with more agency where we get to choose. What I love about these digital sovereign states is you can just pick up your assets, sell them, and leave and move to the next one. It's really hard for you to leave the United States. My wife is a US citizen, and we live in the Cayman Islands. It's a nightmare because US tax and US jurisdiction, because the US is actually a very unfree place once you travel.

13:13You can't open a bank account, you can't do anything. It's actually very difficult. But these digital sovereign states, you can just sell it. You just move on and move on to the next one. I don't like Bored Ape, I want to join CryptoPunks. I don't like Ethereum, I want to join Solana. I like Bitcoin, I want to just own Bitcoin. It becomes very, very, very interesting. So I think there's a solve there. So let's go to the broader one, which is society itself. Society is going to struggle with exponential age, not just AI. It's going to really struggle to understand what is truth, what is my role, who humans are, and where is this all leading to?

13:56Because if you put unlimited energy at a low price, virtually zero, you add AI, robotics, and all of this stuff, You add networks from space outside of sovereign control, which is what's happening. I mean, people don't really understand space is a non-sovereign place of which there is a massive commercialization going on on a scale most people don't understand. We've got all of this kind of megatrends going on. And for society itself, I don't know how we can keep up and how we do not draw the dot, dot, dot, add quantum computing end of humanity. I cannot get there where we don't have that. Because once you put exponential computing power with exponential knowledge, with which self learns, with zero cost of energy, I mean, humans are pointless.

14:48That's very hard. I think that there's one of the reasons that I'm so attracted to this conversation. The stakes are so high, obviously. But it's also, of any technology that I think any of us have ever experienced, the easiest to both see the utopia and the literal sort of end of humanity. So on the utopia side, just weaving together all the threads of what you just said, if we successfully make it through the transition, which there's no way for it not to be brutal and difficult, there are pretty strong arguments that humans that don't feel their value is strictly based on what they contribute to a random corporation or economy your job is a better thing, right?

15:30There are arguments that are pretty compelling that when everyone can code, sure, maybe it makes the salaries that can be commanded by the average computer programmer less, but the things that people can build become so transformationally different, right? And I would argue that you are seeing little glimpses of that already. I mean, it is an overused phrase, but a Cambrian explosion of people building apps already in this AI ecosystem that are, in many cases, them literally using chat GPT to do the coding without having coding experience themselves and sort of figuring out how to wire it all together.

16:10So you have all this sort of optimistic stuff that's both, you know, maybe we get to this point where people have a different sense of purpose and meaning, and the natural tendency of humans to fill vacuums with more creation of all different types comes through to the fore. Now, the flip side is, of course, I think, one, the risks in the short term, the utter displacement, chaos, you know, leaving behind of the people who are unable to make that transition. But then two, the actual discussions of safety questions, existential risk that come up from AI. How do you view the safety conversation?

16:52What's your perception of the AI safety conversation right now? RAOUL PAL Okay, there's a lot of parts of that I just want to get to before we get to that. One is, yes, it's very disruptive now. This job cycle coming out of this recession is going to be very slow, because a lot of people won't come back into the labor force, which is that whole argument for more quantitative easing. So I think that is a very real moment in time. The other side of this is a renaissance, where humans are empowered in ways that they've never been empowered before. And we will see some unbelievable things. We will see unbelievable breakthroughs in medicines.

17:33We'll see unbelievable breakthrough in healthcare, in longevity spans, all sorts of incredible things. But then we've got the existential risk. We've referred to, it's basically the singularity. Do we merge with the robots? Do the robots overrun us? But there's some other existential risk. I spoke to somebody at Google X, and he's like, well, we're not worried about that. It's like, that's further down the track. I said, what are you worried about? Because we're terrified that everybody's actually releasing the technology slower than they've actually got. And he's like, well, what we're worried about is genetic gene editing using AI.

18:12Because you can do rapid scientific tests, which we've been unable to do. You don't need the hypothesis testing scientific method. And he said, we're worried that you can edit genes and create a virus that says anybody with blue eyes should be exterminated. I'm like, really? He's like, yeah. I hadn't even thought about that. These are the things they're grappling with. The other larger part of this is people hate this, but I know it to be true. Well, I think it to be true is that, OK, we've got the shrinking population that's not going to change. In fact, the global population peaks out in 2100, I think.

18:50And the world will shrink. All the Western world is shrinking. And that seems great, less competition for resources. And you hear Elon saying this thing, which is like, the biggest problem humanity faces is less people. First order thinking is, well, why is that a problem? Because we have more resources. Second order thinking is, oh, well, GDP is driven by demographic growth. So GDP growth is slow. Third-order thinking is the way you deal with a shrinking global population is robots and AI. So remember, we talked about these unintended consequences of ways out. The unintended consequence of all of this is it accelerates the use of AI because we've got this productivity need that we talked about.

19:37So this is why he wants to go to Mars. It's driven by having another way out for humanity. And it sounds ludicrous, but if you follow the chain of thought, it's dead right. A collapse in population growth in a indebted world leads to a need for productivity. So that means that governments get behind productivity that we're seeing, and that leads to an acceleration in technology. If you've got less people, you replace them with robots and AI. Problem is, is robots and AI are exponential in their knowledge scaling. And before you know it, you've killed humanity, in which case you need to go somewhere else.

20:16So anyway, that's, there's a lot of steps and phases to this. And I can't remember what your... No, that's, I mean, so what I asked in short was, you know, your perception of the AI safety conversation. And that's, I think, a dimension of it is that when you start to see these patterns, they're less disconnected than they seem. Elon is not doing a random set of activities that are all distinct from one another, but actually have a certain logic. And you can disagree with the logic, you can disagree with the sort of assumption set, but they are coherent with one another, let's say. And I think that one point to hang on for a moment that is super salient is, is this idea that the AI safety conversation probably needs to include adversarial use of the technology where it already stands right now to say nothing of, you know, further developments on the path to advanced general intelligence.

21:08You've got 0.0 chance. This is not the atom, which requires state financing and funding. Even that's not clear anymore. Now, Elon just sent the biggest rocket in all existence into space with private funding. You can limit the amount of uranium that people can process with nuclear technology. This is everywhere and anywhere. I don't know how you stop it. What do you do, shut down the internet? It's like crypto, right? It's like, it's a cockroach. You can't kill it. And I don't think you can kill the AI. And I think this ludicrous idea of alignment, we must get them aligned with humanity. But then when you listen to Sam Altman speak, and you say, well, how does this work?

21:52He's like, I don't know. When you listen to him speak to Lex Fridman, he actually asked the honest question, which is, do you think this is AGI? I don't know what it is. And it appears to learn, and it appears not to learn in human ways, which was the same observation DeepMind had when they saw it play Go. And if you watch the documentary, at first it was predictable, then it lost a game, then it became completely unpredictable. And it won every single game ever, because it never played another human move again, essentially. So I don't know how you align it. Humans aren't aligned. This is Ahmad Mostak's big point.

22:30He's like, I'm sorry, but Pakistan AI is different to Indian AI, because there's no alignment of the philosophical parts of society, the memes that run their lives, religion, societal construct. The US cannot be the AI for the world, because it doesn't fit China, or the Philippines or Japan? Why should it? I don't think there's alignment because humans aren't aligned, except they don't want their mutual destruction, but they still seem to want to kill each other all day, every day anyway. And I don't think you can regulate it. And I think Emad's move was genius with stability AI, which is open source, and also terrifying.

23:13Because he said, this is the most powerful technology the world will ever invent, and And we cannot just give it to Google and Microsoft and let that run it for the world. And then maybe a Chinese version. He said, that's not right, because that's the mass destruction of global culture, let alone it's worse than the Middle East and the US having all the oil. That causes enough problem. This would be a catastrophic situation. So he's like, so open source it. And India can have its own AI. Pakistan can have its own AI. The Middle East. everybody can have their own culturally relevant AI, et cetera.

23:50Brilliant. Makes it completely unstoppable, which is terrifying because it accelerates it. So I don't know the answer. I think there are no good answers here. And so my argument with all of this is we don't know. It's clear there's a lot of outcomes of which many of them are terrifying. I'm just going to blindly say, I'm going to throw myself into this and buy the ticket, take the ride, see where the goes. I will reveal my hand a little bit. So, you know, I think that the, so I'm very similar to you in the sense of, uh, there are things that feel forces more powerful than our ability to sort of stop and slow things down and et cetera, et cetera.

24:33Um, I think that the conversation about AI safety is enormously important. And I think that you have a hundred million new people who now have, who understand they have a stake in that after experiencing ChatGPT and working through it for the first time. And the landscape of that discussion that they're coming into is doom prophets and Shogoth memes on the one hand, which are completely inaccessible, largely driven by people who feel like they uncovered some secret of the world before and are almost condescending to everyone who didn't figure it out before. So that's one side. The other side are the accelerationists who are just excited to race to Terminator mode.

25:18It feels like there's probably a real big space in the middle for those folks who are genuinely, you know, filled with wonder when they, you know, take some idea from their four-year-old kid and turn it into an image on Mid Journey, but who also are seeing that there's huge implications. And I think it's important to have a different space for it. So I kind of like your take of diving in, but at least having that conversation. I do think also that the open source piece is really important. I do. But I also see what you say. And maybe we're going to have to self-opt in into restrictions. Because I don't know how it can be done at governmental level.

25:56So we're going to have to do it at distributed nation state level. So I align better with XY AI and not the other one. I don't know the answer because it's like crypto. It's impossible. The The Chinese ban it, nothing happens. The Indians ban it, nothing happens. The US stamps on it, nothing happened. I mean, it's unstoppable because it's distributed. That's the whole point of it. And it's the same with AI. So I get your point, and I agree with it. It's a really important conversation. But who's going to have the conversation with whom and how? And the tech companies, when you see Sam Altman, he's terrified.

26:34I asked him, I said, where's this going? His answer, I said, where's this going to be in five years? He just laughed at me and said, mate, I didn't think we'd get chat GPT-4 until the end of this year. He said it came out in two months. He said, we have no understanding. And then you think that Microsoft don't seem to care. They've just seen a business opportunity so big that they can destroy their competitors, and they're just pressing the nuke button. I mean, you're putting chat GPT-4, GPT-4 overall, in all Office? I don't know how many Office users there are, a billion? really? And nobody's going to stop them because they can do it tomorrow.

27:15Well, they're already doing it. And then every Bing browser and then Google are going to be forced to do it. They're going to be forced into the Google browser and then G Suite and everything. And before you know it, we've got 5 billion AI users. How the hell do we stop that? Yeah. I mean, it's already happening. I mean, the number of people who have switched over to Bing, you know, you see the New York Times just reported the other day that Google started losing its shit because Samsung came to them a few months ago and were like, we're actively considering removing Google as the default browser and putting Bing in.