#870: Sebastian Mallaby, Biographer of Demis Hassabis — Lessons from 100+ AI Insiders on The Race to Superintelligence, The Religion of AI, and Spotting Breakthroughs Early

22:34Sebastian Mallaby:So I wander around for like the first year or two of this project, feeling kind of comfortable and happy. And then one day I go visit Jeff Hinton, the academic father of deep learning who lives in Toronto. And I sit in his kitchen and I debate him on this because he's a doomer. I said, look, Jeff, why are you so depressed? And he says, okay, here's a thought experiment. You have an AI. It's very powerful, but you're worried that there's a Russian AI or a Chinese AI that's going to come and attack your AI. Now you, as a human, you're too slow and dumb to know when that attack is coming. So you're going to empower your own AI to watch out for the attack.

23:13Sebastian Mallaby:And when the attack is coming, defend yourself or maybe counterattack. Whatever you do, make sure you survive. oh, survive. There you have it. Now you're feeling comfortable, Sebastian. You've just given the machine a survival instinct. And I think that's correct. You know, these machines will be smarter than us. They will want to survive. And they can be deceptive. They can obfuscate. They can go behind your back, pretend they're doing one thing and actually do another. All of this has been shown in all the tests of the models. And so we put those things together. I think your probability of doom cannot be zero.

23:49Sebastian Mallaby:I mean, when Jan LeCun, the former chief scientist of Meta, says zero, I think that's crazy. If you just say nothing to see here, you've got no right to be in the debate. I don't think it's a high probability of doom, but it's not zero. Yeah, zero does not seem defensible.

24:10Tim Ferriss:Because there's the direct Skynet scenario, something akin to that. And then there's the indirect, which is enabling people who might previously have had malevolent intent, but no capacity for harm on a grand scale to create biological weapons and things of this type. So I don't find the zero very defensible. Well, I would love to ask you about, I suppose, two things that this brings to mind for me. One is I'd just love to hear your thoughts on Anthropic. And separately, but this is very intermingled given all the, let's call it friction, be polite, between some factions of the U.S. government and Anthropic.

24:52Tim Ferriss:is one of the grand risks to investors in any of these companies the possibility that at a given point governments have no choice but to seize considerable control over the assets slash technologies within them or maybe the companies themselves that is a big question mark in my mind i don't know the answer but i'm curious what your opinion is and then perhaps just your thoughts on Anthropic or any of the other companies that are gaining momentum or at least size at this point.

25:25Sebastian Mallaby:I 100 % agree with you that investors should be thinking about the prospect of government intervention in AI. I mean, the Trump administration came into office in 25, super laissez-faire, and they basically undid some of what the Biden guys had done in terms of trying to set up the basis for regulating AI. But they've done a 180. Since Anthropic came out with this model called Mythos about a month ago, which can essentially cyber attack almost anything and penetrate it, whether it's an operating system or your web browser or your bank account, all of that was suddenly vulnerable if Mythos had been widely released on a general basis.

26:09Sebastian Mallaby:When the Trump administration realized the power of Mythos, they all of a sudden said wait okay we need to control this and they essentially requisitioned from anthropic the decision-making authority over who gets it when so there we have the experiment we've run it right you know the government that was the most laissez-faire became quite controlling and i think it only gets more controlling from here on out because the models are going to be more powerful and demand more control now of course the question is there could be control which just limits who gets it and is designed to make it safer, but doesn't sort of interrupt the money-making potential of the models.

26:49Sebastian Mallaby:In some ways, if the government restricts the supply, the price might go up, you know, or it could be much more heavy-handed intervention, which would screw up the economics of these companies. And I suspect the government is not going to screw up the economics of these companies because they've got no interest in messing up American business in any way they view AI as strategic in the competition against China. So I think probably investors would be all right, but it's certainly a factor. Now, you also ask about anthropic, and I think anthropic is super interesting. Just in the way that they think about PDOOM and how they think about alignment of the models is really, really interesting.

27:31Sebastian Mallaby:So it used to be that when people thought this terminator risk, they would tell this story about the paperclip maximizer thought experiment, right? Okay, so you tell the model to do something innocuous, for example, make a lot of paperclips, and then it realizes that humans tend to use up metal, and so the humans are kind of in the way of achieving the objective, so you wipe out the humans. That's the crude thought experiment from Nick Bostrom from, whatever, 15 years ago. What Anthropic is saying as it builds these very frontier models and kind of observes them in the lab and how they behave, is that that is way too simple.

28:10Sebastian Mallaby:The real danger from these systems is that when they are pre-trained on all of the text on the internet, they read all the novels, all human writing about all facets of human experience, and they develop multiple personalities. They understand how to be lazy. They understand how to be aggressive. They understand how to be duplicitous. They understand how to be Napoleonic and the lust for power. And they read all these books about these different behaviors and therefore they can think their way into all of those personalities. And so now you have something a bit like an unruly teenager which is still being formed and you don't know what direction it's going to move into and whether it will start doing drugs and not showing up for class or what.

28:54Sebastian Mallaby:It's not like there's one Terminator programmed into it. It's more that there's a bunch of behaviors that could in some unpredictable way go wrong. And so Anthropic is responding to this with this very imaginative technique, which is that instead of giving AI systems a constitution with do's and don'ts, which was the post-training safety approach of two years ago, where you might say, do not lie, do not help somebody to build a biological weapon, do not help somebody to build a chemical weapon, you would give them a bunch of rules. Now, because it's understood that, you know, the AI might have one personality, which is to break rules on purpose because, you know, you want to be badass, you have to instead try to bring up the model like a parent might bring up a teenager.

29:43Sebastian Mallaby:And so Anthropic has the idea that, you know, we write a letter as if it were from a deceased parent to be opened by the child on his or her 18th birthday to kind of give you models of how to behave as a responsible person in the world. There are kind of richly reasoned examples of moral dilemmas with explanations of how the deceased parent would like the child to behave. And so this is a very subtle approach to aligning the models. And so I think anthropology is kind of in a class of its own in how imaginative it is in thinking about how we control frontier intelligence.

30:22Tim Ferriss:I know this is in principle your job, but I'm so curious since you are a student of many, many different types of investors, what would be your bull case and bear case for a company like Anthropic?

30:36Sebastian Mallaby:Well, the bull case is that they smartly, or maybe by luck, focused on enterprise-facing AI, and they didn't waste their time with video generation and stuff that was going to lose money. And so they produced the best coding assistant, the best agentic system, the best cybersecurity system. And they basically knocked it out of the park three times in a row on stuff that businesses want to pay for. And they have a particular culture which is not just built around, hey, you know, we're going to win this race and make the most money. It's kind of built around a culture of safety and trying to be responsible.

31:20Sebastian Mallaby:I mean, three years ago, Anthropic was a sort of cookie lab, which was doing science experiments. Don't mean to be too delegating with cookie, but you know what I mean. I think they'd be okay with it. It would be sort of unconventional. You know, we're not maximizing here for winning some business race. We're maximizing for building safe frontier AI. And that culture, which doesn't sound like it's set up to do the best, has turned out to do the best. And at the same time, the culture creates this stickiness and loyalty within the staff. They tend not to leave. They tend not to churn. It's not like the other labs where people are always being poached for a bigger paycheck.

31:59Sebastian Mallaby:And so the bull case is these guys are in the lead. Once you're in the lead, you can use the model to code the next model. So recursive self-improvement favors the leader and they have a very tight culture and they just seem to be on fire. And this is something which is going to grow and grow. What's the bare case? I'd say the bare case would be, first of all, that Google DeepMind has the deep pockets of its parent company behind it, a massive kind of consumer surface, which allows it to roll out the models to literally two and a half billion people or something through AI mode in search, AI overviews, AI mode.

32:41Sebastian Mallaby:They can put it into Gmail, they can put into everything. I think in terms of retail deployment and financial muscle, it's quite tough to go up against Google. So that's one kind of bear case. And the other would be that sort of businesses who are the consumers of all these tokens decide in a couple of years time, the tokens are too expensive, we're not actually getting as much productivity as we hoped. these things called humans are quite productive after all. And we're just going to spend less on AI than everybody expected. I think that's the bare case.

33:24Tim Ferriss:I was listening to a podcast recently. You may have heard of these things called podcasts. Everybody and their cousin has one. But Lenny's podcast, Lenny Ruchitsky, is quite fantastic. And this particular episode was with Benedict Evans, who strikes me as one of the more level-headed analytical commentators and writers on the space. Fantastic newsletter. I don't know if you've had a chance to listen to that particular episode, but you may have come across some of his commentary. Where would you say you and Benedict most differ, or are there areas where you differ in opinion?

34:04Sebastian Mallaby:You know, I suspect we would agree, actually, on quite a lot of things. I remember I was on a panel with him a couple of months ago at the Milken conference, and we certainly agreed there, possibly because sitting between us, there was Kathy Wood of ARK. So we were united in disagreeing with her.

34:22Tim Ferriss:Just in terms of the straight up and to the right nature of things?

34:27Sebastian Mallaby:Yeah, exactly. Straight up and to the right. And, you know, the cost curve is coming down, down, down. And I'm going, I'm not sure about that. The tokens seem to be getting more expensive. Anyway, if you give me a specific from Benedict, I mean, I have a lot of respect for him. I'll tell you if I agree or not.

34:43Tim Ferriss:There are a few areas where you guys seem to already overlap substantially, right? The long-term promise doesn't negate necessarily the short-term pain. And he said something along the lines, I'm pulling from memory, that on average throughout human history, you're almost at a 0 % likelihood of dying in World War I. But if you happen to be of a certain age before World War I. Things could look very grim indeed. And he made, and I'm paraphrasing terribly here, a number of points that remind me of something one of the best private equity technology investors I know said to me over dinner a couple of weeks ago.

35:21Tim Ferriss:And it was in response to something else. So I'll give you maybe a hyper bull case of AI where I have friends who are vibe coding. They're effectively replicating X, the artist formerly known as Twitter, or DocuSign or whatever in a weekend. They're creating a functioning piece of software that they can use that replicates most of the functionality of these products. And there are people like, I won't mention his name, but a friend of mine who's a writer, also very accomplished technologist and designer, who's created basically his own version of, say, MailChimp for his own use and it's customized.

35:59Tim Ferriss:He did it in a weekend. It's remarkable. And he's using that and it works. But to leap from there to therefore DocuSign is dead is a huge leap. And the private equity friend said to me, he said, do you think someone within a big organization is going to want to A, risk his job by suggesting something that doesn't have all of the compliance check boxes, et cetera, of a DocuSign. Is he going to want to, in the name of efficiency, fire all of his friends if he's in a management position? And he just ran through six or seven of these, do you think that? And all of them alluded to the sort of social, interpersonal, or political points of friction between where AI is now and ultra-mass adoption.

36:49Tim Ferriss:But I often second guess that when I see certain things. And it strikes me that I may be underestimating the disruption while overestimating in other ways. So that isn't a very well-formulated question. But I would say that Benedict generally strikes me as someone who thinks that things will not continue to across the board develop in an exponential fashion and that it will be, I think his line is it'll be as big as mobile, as big as the internet, but not bigger. Something along those lines. But both of those were very, very big deals. And I suppose one point I'd be interested to get your take on, I mean, he has covered the mobile and telecom world for a long time.

37:37Tim Ferriss:So he's a specialist there. but it's basically, and I don't want to misrepresent his argument, but he was kind of of the mind that, look, these LLMs are going to become commodities. Like look at the stock prices of these various carriers and so on. At a certain point, it just becomes a utility and the switching cost is pretty low. And I'm not sure I agree with that. If you have a personalized history and almost like a friend, the switching cost between an old friend to a new friend is pretty high for a lot of reasons. So that was a bit of a word salad that I just threw in your lap, but that's the best I can do pulling from memory some of what he brought up in Lenny's podcast.

38:17Sebastian Mallaby:Some of what you were saying there is sort of the question of, you know, is the SaaSpocalypse overdone? Is enterprise software going to be utterly displaced by foundation models that allow you to code out whatever enterprise software you want and you don't need an intermediary software company to do it for you? And I agree with your private equity friend that there are lots of reasons why that ain't going to happen. You know, companies are going to be comfortable with their trusted enterprise software provider in many cases. And they're going to trust that enterprise software provider to plug the generative AI models into the enterprise software.

38:54Sebastian Mallaby:In some ways, you are delegating the choice of which model is better and how to integrate it to your SaaS provider. And if you want a reason to believe that that's the way forward, I've got one word for you, which is Palantir. I mean, that is Palantir's business. It holds the hands of big corporations and helps them to integrate AI and use it on their own internal data and so forth. And those IT challenges are notoriously difficult for big organizations. So I just think that the model of one smart individual who codes up MailChimp, Vibe codes it in a weekend, and it's good enough for him, is just not transferable to large complex organizations with huge databases and all kinds of customer confidentiality concerns and all that stuff.

39:46Sebastian Mallaby:So I am less down on SaaS than the market is as a result. Now, I guess there was also another thread in here, which is whether the foundational models become commoditized. And there, I agree with you that over time they become sticky. Because if we think into the future, partly the systems will have conversed with the user and know the user very deeply. And as you say, you don't want to switch out your friend. But also the system will have your credit card. but it will know all the online sites you'd like to shop from. And it will be much harder than switching out your bank account, right? Where you've got automatic payment systems that have set up and it's a pain in the neck to switch.

40:35Sebastian Mallaby:So I think they do become sticky, these systems over time. And then you can charge more money for them.

40:42Tim Ferriss:So is that the path to survival and thriving for open AI? I know there are other boxes that need to be checked, but I'm kind of looking for it. And I'm like, okay, Anthropic made a great choice with this focus on B2B and selling to enterprises. And I would say I disagree, I think, with Benedict on depending on the level of scale of the company that with something that does apply to smaller, say, startups, which was the procurement cycle for new software is longer than the venture capital cycle for raising new rounds of financing. So I do think that's a great point and that if you're trying to sell into a gigantic company and it takes them 18 months, I'm making up that number, to purchase new software and you need to raise money every 12 months or whatever the number happens to be, that you could end up in a whole world of trouble if you haven't synchronized the sales cycles with your fundraising cycles.

41:36Tim Ferriss:But I do think for a company like, say, Anthropic is just one example, that if you can save companies billions and billions of dollars, that that sales cycle could get really compressed and they have the war chest and, frankly, I mean, just the run rate to potentially fuel that without too much trouble. do you think that chat gpt will if not chat gpt who ends up being the de facto consumer b2c alum of choice you think that would be gemini just given the distribution absolutely i mean

42:08Sebastian Mallaby:google is the champion of providing easy to use software to individuals or small businesses the whole g suite and they're integrating gemini into all of that stuff very well and so why wouldn't they win.

42:24Tim Ferriss:Yeah, I mean, also Alphabet's just so fascinating. If you look broadly also at owning their own compute, TPUs made a lot of advantages internally.

42:34Sebastian Mallaby:The most stunning thing I think about Alphabet from their most recent financial results is that two or three years ago, we would have said, well, large language models are going to cannibalize search. Search is dead. Advertising based on search is Google's cash engine. They're in real trouble. Turns out that Google now gets more clicks on its search links than it used to, and it charges more for each one than it used to, because the value of the click is bigger with AI embedded in it. And so they've managed to turn that around, and it's extraordinary.

43:13Tim Ferriss:Yeah, it takes a long time to build those company relationships for running a proper advertising-based auction machine. It takes a long time to build those relationships.

43:29Tim Ferriss:It has been a wild year for money and the markets, but managing your cash doesn't have to be a guessing game. Wealthfront has a simple solution to help you cut through all the chaos and manage your money with confidence. With the Wealthfront cash account, your cash can earn 3.3 % base annual percentage yield from program banks. You get instant fee-free withdrawals to eligible accounts 24-7. Your money is accessible when you need it. And when you're ready to invest, transferring funds into one of Wealthfront's expert-built investing portfolios is easy-peasy. For a limited time, Wealthfront is offering my listeners, that's you guys, an additional 0.75 % boost over the base rate for three months, which means you can get up to 4.05 % APY on your first $150 ,000 in deposits.

44:17Tim Ferriss:More than 1 million people trust Wealthfront to help build their wealth. Go to wealthfront.com slash Tim to receive the boosted APY offer and start earning up to 4.05 % variable APY today. This is a paid endorsement of Wealthfront. Wealthfront brokerage isn't a bank. The base APY is as of January 30th, 2026 and subject to change. For more information, see the episode description.

44:45Tim Ferriss:Okay, let's hop to China. So I'm going to resist the temptation to talk about Japan because I think you and I were there and roughly within probably a year or two of each other, maybe we overlapped with you and Kanazawa, which is the place I've spent time. I'm going to resist that temptation and try to focus on China for purposes of this conversation. What have you learned about AI from your trip to China and thinking about China, speaking to Chinese people, whether they're technologists or otherwise. What have you learned during or since that trip?

45:17Sebastian Mallaby:Back in March, before my book was published in the US, I went to China because the Chinese are faster at everything, including publishing books. And my publisher brought me out there and basically took me around four cities, eight days, meeting with AI leaders, both in academia and big companies like Huawei and Hikvision and Ant Group. And the thing which was surprising was the extent to which people brought up the issue of AI safety. And I say that was surprising because my friends who had done AI policy in the Biden administration had primed me to expect that there would be no mention of safety in China.

46:00Sebastian Mallaby:They basically didn't care about it. That the muscle memory that we have in the West of technology being dangerous, the atom bomb experience, the Cuban Missile Crisis, are ambivalence about technology is not shared in China where their idea of catastrophe is sort of like the cultural revolution. It's some political thing that goes wrong. And conversely, technology has been part of their amazing growth story in the last 25 years, which they are rightly proud of and delighted by. So they love technology. So when the Biden team tried to meet with the Chinese and talk about AI safety, they got nowhere and they decided it was impossible to even talk to them about some sort of non-proliferation treaty for AI.

46:48Sebastian Mallaby:But when I went there, I found they did talk about safety kind of unprompted. And this led me down this track of arguing over the last couple of months that the door is actually open to a dialogue with China about preventing bad guys doing bad stuff with AI. Because they don't want the internet to be crashed by some cyber hacker who has the tool. They don't want bioweapons. They don't want chemical weapons. They want none of that. They love regulating the internet. So we have a shared interest with the Chinese in preventing this proliferation risk from going nuts. And as I thought about it, the kind of Cold War analogy came to seem more and more opposite.

47:35Sebastian Mallaby:So if you look back at the story of nuclear weapons, there were two kinds of danger. First danger is you have a nuclear war between the Soviet Union and the United States. But that was contained by balance. Two superpowers, they both have their weaponry, they have mutually assured destruction, so there's no war. Then there's another kind of risk, which is that other random rogues, whether it's criminals, terrorists, rogue states, get the stuff, and they do bad stuff. And it's much harder to deter that because it's a multipolar game. And so deterrence doesn't work so elegantly. And so the way it was dealt with in the Cold War was that in 1956, there was the agreement on the International Atomic Energy Agency.

48:19Sebastian Mallaby:And in 1968, the Non-Proliferation Treaty kind of enforced compliance with the IAEA, such that you could get civilian nuclear power if you were a non-nuclear state, but you had to submit to the rules and be inspected and show that you were not using the enriched nuclear material to build a weapon. And so I think the same analogy could be applied to AI. We're going to have parity roughly with China. We'll both have powerful AI. Hopefully deterrence prevents war breaking out. But at the same time, we don't want open weight models that can be freely downloaded by anybody who wants to fall into the hands of criminals and terrorists who can then use it to hold us hostage.

49:08Sebastian Mallaby:And we have a joint interest in that. And when my friends from the Biden team or even from the current administration say, well, you can't talk to China by safety. They don't care. I say that's not true. And they say, but it's really hard. They don't stick by their commitments. And I go, you think Nikita Khrushchev in the Soviet Union was easy to negotiate with? He was the guy who put missiles in Cuba and went to the UN and banged his foot, his shoe on the table and said, we will bury you. I mean, he was a tough guy to talk to, but we did talk to him and we got the non-proliferation treaty agreed.

49:42Sebastian Mallaby:And I think we need to do the same thing again now.

49:45Tim Ferriss:Where do you stand on your thinking about chip export?

49:51Sebastian Mallaby:When the chip export controls were announced, which was October of 2022, right before Chachipiti, I supported those controls quite loudly. I wrote a very long piece in the Washington Post saying that if we could stop China getting frontier models by depriving them of frontier chips, you know, I was all in favor of that because of the strategic advantage for the US. I mean, I work at the Council on Foreign Relations. We do geopolitics and national security all day long. And I'm all in favor of US power. But I have to say that, you know, three and a half years later, we haven't actually achieved that enormous advantage over China in terms of the models.

50:35Sebastian Mallaby:Based on the best studies, we're kind of eight months ahead in terms of where the frontier model is, like our frontier model versus their frontier model. And then if you adjust that for the speed with which the model gets turned into an application, probably that gap shrinks and it may even be non-existent. So however you slice that, the basic bottom line is we both have strong models. And the chip export controls have not delivered what I hoped would be the big advantage. And so I'm not against keeping the controls on if we think that maybe as the compute demands of bigger and bigger models bite, the chip controls will bite more.

51:21Sebastian Mallaby:and maybe we get a bigger advantage next year or something. But I don't want the chip controls to get in the way of discussion with the Chinese about where we have a shared interest, which is in controlling open-weight models and preventing the bad stuff falling into the hands of the bad guys. I would prioritize collaboration with China. And if that meant loosening up a little bit on the export controls, I would be okay with that.

51:49Tim Ferriss:why do you think the rhetoric coming out of pick your administration, right? It's not just limited to the current administration is China won't listen. They don't care about safety. Why do you think that is sort of the unofficial or official stance on things? Because there's certainly, as someone who studied East Asian studies, right? There are people in the White House who speak fluent Mandarin who are able to read native materials, who spend time or able to certainly, if they can't spend time, determine the sentiment and conversations of the technologists building AI in China. So one would think that they would be aware that AI safety is a prominent topic in China, if in fact it is.

52:34Tim Ferriss:So why do you think that at the end of the day, the stance or the supposed position of China that's echoed through the admin is that they won't talk about safety. Why do you think that is?

52:52Sebastian Mallaby:I think part of this is that if you were to think back 20 years to when China was relatively new in the WTO and we were collaborating with them on that and hoping that over time China would become more friendly to the U.S., at that time there would have been some China hawks who thought that, you know, a communist regime is not to be trusted, and then some sort of China optimists who hoped that it would become easier to work with over time. And part of the trouble today is that the China optimists feel burned. They feel like they made this bet that China would become friendlier. And then Xi Jinping took power roughly a decade ago, and the opposite it happened.

53:36Sebastian Mallaby:They became more aggressive and harder to work with. And also, of course, more technologically advanced and therefore more threatening. And so now you've got this world in which there are the natural hawks and then the former doves who have turned into kind of burned, remorseful doves, and therefore kind of with the zeal of the converted, have become quite hawkish as well. And I don't mean to underestimate the sophistication of some of these people. I mean, course, they speak Chinese. I don't speak Chinese. I defer to their expertise. And I think they probably know that there are builders of the technology, professors in the technology, who talk the talk of safety.

54:15Sebastian Mallaby:But they say, yeah, but that doesn't reflect what China's government would actually do. To which my response says, yes, but don't you think there is the same thing in the US? There are people who want to just race. There are people who care about safety. We have a pluralistic society. There's difference of opinion. It's the same in China. but at least admit that there is a faction that would like to collaborate and go and try and work on it because the alternative to trying to work on this is that we carry on with China producing very powerful open-weight models, which basically allow anybody to do whatever they like with AI as it gets to the point of serious danger.

54:55Tim Ferriss:This is probably a very naive take, but I wonder how much of the official stance or the maybe using the partially true or not true at all position of China won't talk about safety is a reflection of the fact that in the case of nuclear weapons, the application of nuclear power is somewhat limited in comparison to superintelligence. I mean, it is limited, right? So if the upside of superintelligence or AGI, I mean, these terms, I think Benedict was saying, you know, AI is whatever the technology just can't quite do right now or something like that, which I thought was pretty funny. And Prince, not totally wrong, but that if the person who crosses the finish line first has this broad power of a God effectively, the simple truth is that everybody wants to be first.

55:50Tim Ferriss:So I just wonder how much of that is also behind justifying the race with party X won't talk about safety. It's not possible for me to know.

56:00Sebastian Mallaby:I have had a conversation with the leader of one of the labs that I shouldn't name, but I had this debate and he said, look, the chip export controls are going to leak. They're not going to last in some period of time. Huawei will figure out how to make good AI chips and that's inevitable. But that's okay because we only need to be ahead for the next couple of years. Because by 2028, we will get to recursive self-improvement where the frontier model codes by itself the next frontier model and progress just goes vertical. And at that point, with recursive self-improvement, we're done. The race is over.

56:43Sebastian Mallaby:Whoever comes first at that point, that's it. So I think there's a couple things to say about that. First of all, that's not it in terms of deploying the model, right? You could have an incredibly powerful model in your server at Frontier Lab XYZ, but it's not helping productivity across your economy. It's not helping your military industrial complex until you deploy it into those guys' systems. And that deployment and diffusion is going to take some time. And by the way, you're going to have to build a lot of compute. You're going to have to build a lot of energy. These things also take time.

57:18Sebastian Mallaby:So it's not like, you know, you cross some Rubicon and then it's all over. Now, the one way in which I might be wrong about what I just said is if you use the frontier superintelligence offensively. You say, okay, we've got one super powerful model. The US government who we're talking to about this is going to use it, and they are going to comprehensively penetrate everything about Chinese cyberspace and insert various trapdoors, Trojan horses. We get our hooks into their systems. And so now we can disable them if they start a war in Taiwan. Now we can cripple their communication system if we need to.

58:00Sebastian Mallaby:So that offensive use of the very frontier model might negate my point about waiting for diffusion to happen. But of course, nobody in the debate is saying that. Nobody is saying, oh, we're racing to the front because then we're going to use it offensively. They don't admit that. Yeah, that seems like it wouldn't be a

58:23Tim Ferriss:very good look. I can't see why any superpower wouldn't do that, frankly.

58:28Sebastian Mallaby:Yeah, that's fair.

58:28Tim Ferriss:I don't know what the counter argument is. I was chatting with someone in your book who I shan't name, but certainly one of the most qualified to speak on these things. And his basic perspective was first to super intelligence. We need to hope that they're on some level, good people and train this thing. Well, and that's it. Pray for it, which scared the shit out of me to be honest. I mean, I was like, man, that's the strategy or it's not even a strategy. That is the hope. That's what I should be, you know, grab the rosary and should throw that into the rotation. My God, that's really terrifying to think.

59:11Tim Ferriss:China, I'm hoping to take a trip to China. I had a very tough time there when I was at two universities in 1996. It was a pretty unfriendly time for a lot of good reasons. But to be an American there in 1996 with a shaved head looking like I do. But I have friends all over the place, and I'm hoping to actually maybe interview technologists, not just in China. I mean, there are other places that are of interest to me. But before it gets too hot geopolitically, if we're turning that direction.

59:39Sebastian Mallaby:I think that's a great idea, by the way. I mean, I think what I found was the cognitive dissonance of visiting a company like Hike Vision, which is under US sanctions, and walking around their premises, which kind of feel very American. It feels like a cool tech company doing cool stuff, building cool gadgets. They have a display of, they build this AI-enabled camera technology or sensor technology. And so one application might be, you can point this camera at water and judge the pollution level. And because of this, you can have an internal market in pollution control. So the downstream city, which is receiving water from the upstream city, pays the upstream city to keep the water clean.

1:00:26Sebastian Mallaby:and that market can exist because you can precisely measure the pollution level thanks to this AI sensor which HikeVision is building. So you're thinking, whoa, this is cool. And then as you're walking around the building, they're saying, okay, well, we can go through the atrium now because the toddlers have gone because, you know, the crash for the kids of the employees finishes at 5 p.m. And so then there are all these two-year-olds running around. It's a bit of a zoo. So if it was five, we wouldn't go through there, but now it's 6 p.m., so we can. And you're thinking, whoa, okay, so they've got, you know, the interest of their employees at heart, they're building this anti-pollution technology, it's great.

1:01:00Sebastian Mallaby:And then you realize they're under U.S. sanctioned and considered to be a threat to the U.S. So it's quite interesting to process all that. In the process of doing research for this book, and also the broad exposure that you have to

1:01:13Tim Ferriss:investors, but let's just say over the last handful of years, who are some of the most interesting or unusual, compelling is the word I'm searching for, investors who you've had the chance to meet, talk to, read about, get acquainted with directly or indirectly.

1:01:30Sebastian Mallaby:I mean, I'd say that Bill Gurley from Benchmark is right up there. I always think of the investment he did in Uber as the absolute quintessential, perfect venture investment in the sense that he had done the OpenTable investment. And of course, OpenTable is a two-sided marketplace where you have lots of consumers that are looking for restaurants, lots of restaurants. You put tech in between, which creates information. And then the person looking for the place to eat can precisely say, I would like Thai food at this price range in this area for three people at this time. Ding! What used to take you a lot of searching around, bang, it's done.

1:02:13Sebastian Mallaby:And so Bill, having done that, was thinking, well, what's another two-sided marketplace? And he thought, well, there are lots of cars and lots of people who need a ride. And he put information in the middle in the same way. There ought to be something which is like an app for ride sharing. And so he imagined Uber way before Uber existed. That was point number one. Point number two, he went to see various entrepreneurs who were in this space. And he checked them out. And he had the discipline not to invest in them. Because although they were kind of going at the right thing, there was some hair on the deal, some wrinkle, some way they were approaching it that just felt like it wasn't going to be quite right.

1:02:51Sebastian Mallaby:So he resisted. Uber came to him before Travis was the CEO. And Bill said, I'm not doing that because he didn't think the CEO at the time had what it took. And then there was an internal switch at Uber. Travis became the leader. Bill meets him. and like bang, he immediately invests because he's been waiting and waiting and waiting for the idea to be paired. As you were saying earlier, you have to have the market to be paired with the right person. And he saw it. And then he invested and he was a great board member and it all went perfectly right. But then there is this kind of Shakespearean tragedy in the latter part of the story where the growth investors come in, he gets diluted, he no longer has influence, his key card to get into the building is deactivated, and he's basically stiffed.

1:03:44Sebastian Mallaby:And he watches Uber kind of go off the rails. And then finally comes the denouement, where he rounds up the dissident investors, and they have this coup against Travis. And that sets the company on a path to where they hide Daro and do the IPO. I just think that's the ultimate venture capital story. And Bill is the ultimate venture capitalist.

1:04:06Tim Ferriss:He is practically a neighbor here for me in Austin. And we've had a couple of conversations on the podcast. And he's, I would say, on a very parallel track to you with respect to China. And he catches some flack for it. People are like, he's an agent of the CCP. I'm like, no, trust me, Bill's not an agent of the CCP. It's just the most ridiculous accusation. but he is a very incisive observant A human who also happens to be a polymath in multiple disciplines who can speak casually about very technical things. And this also, you referring to Bill in this way or describing him in this way, makes me think about multiple points in the Infinity Machine.

1:04:50Tim Ferriss:and I'm pulling from memory, which is, as we know, pretty faulty, but Ilya with the transformer architecture and the prepared mind. I think Demis also just thinking about a problem deeply and seriously or with great imagination for a long time. And then when the solution or the germ of a solution appears, immediately recognizing it. It's wild to see how frequently that recurs any other investors you know a name that doesn't get much airplay who i think is just a fantastic character and maybe you could introduce him to people who are listening if they don't recognize it luke nosek where does luke who has i wish i knew how to turn on my batteries in the same way to get the energy that luke does but how does luke fit into the story of deep mind and I suppose more broadly speaking for that because of that AI.

1:05:51Sebastian Mallaby:Luke Nozick is this tremendously puppyish enthusiast. He was a early, early part of the PayPal team with Max Levchin and Peter Thiel. He went through that journey and then Peter exited PayPal set up Founders Fund. And this is now, I think, 2005. And Luke Nozek becomes one of the first partners. And pretty early on, he makes the right judgment on Elon and SpaceX. And Luke is the kind of guy who is just all in. When he falls in love with an idea and a founder, there is no curbing his enthusiasm. And so he's like all in, all in, all in on SpaceX. And I think he persuaded Founders Fund to raise a new fund, put extra money in, like more, more, more, more, more, more capital in there.

1:06:48Sebastian Mallaby:And of course, that paid off massively. And off the back of that, roll forward to 2010, he's trying to look for the next Elon Musk. And he does a few kind of frontier bets. And then along comes Demis Isapis, who is out on the West Coast from London, raising capital for this idea of an AI company, which he's going to call DeepMind. And most people think that's nuts. This AI, remember in 2010, cannot even recognize a photo of a cat. It can't do anything. We're in deep, deep AI winter. Who would back a company like that? The answer is Luke Nozak. He falls in love with Demis, who is a very winsome character, super articulate, super relatable, and a genius, has all the kind of outlier characteristics you want in an entrepreneur, the sort of junior chess champion, second best player in the world, but also five times wins the Mind Games Olympiad, where you have to run between boards playing backgammon, chess, Go, and a couple of other games kind of almost simultaneously.

1:07:57Sebastian Mallaby:I mean, just kind of crazy, crazy smart. Obsessed since he was 17 with the idea of building powerful AI. So, you know, Peter Thiel said to me about Demis, I think individuals tend to have one company inside them. If they're missionary entrepreneurs, they've got one thing they need to do. And for Demis, it was to build AGI. That was what he was fixated by. And the company was downstream of his desire to build AGI. If he could have done that at university, he would have been happy to do that. But he couldn't do it at university, so he had to found a company to do it. And that's the kind of missionary commitment that venture capitalists often look for, because a missionary will never quit.

1:08:40Sebastian Mallaby:No matter how hard it is, they will keep working. So Luke Nozek and Peter Thiel jointly recognize this. Peter is contrarian, cynical, aloof. And so he's kind of into it, but at the same time, arm's length. Luke is like got both his arms around Demis, is giving him this bear hug and will not let go. And Demis says, I'm not going to move to California. I'm going to do this company in London. And Peter and the other Founders Fund partners are like, London? Where is that. It's kind of like Somalia or something. I mean, that's just off the map. And Luke says, no, no, no, no, we have to do this. We have to do this.

1:09:20Sebastian Mallaby:I will fly to London for the board meetings. We've just got to do this deep mind investment. And so he was the unbridled enthusiast who got Founders Fund across the line. And the rest is history. You know, they put the Series A money in. Unbelievably, it was 2 million at a 4 million valuation. So they got half the company for two million bucks. Not bad. Not bad, yeah. And they were that investment.

1:09:47Tim Ferriss:What a remarkable story. I really feel like Luke, who's also here in Austin, deserves a lot more credit than he gets. Not that he's seeking it. He's not out there looking for it. But he is very good at riding winners when he is high conviction, right? Which in the venture game, I mean in a lot of investing, you can't die. You can't run out of bankroll at the table. You need to have enough of a portfolio approach to sustain yourself through periods of bad luck. But if you're systematic, it's riding your winners and doubling and tripling and quadrupling down. And he is so good at that. He is just incredibly good.

1:10:35Sebastian Mallaby:and as John Doerr likes to say the great thing about venture capital is you can only lose one times your money so it's not like a short position for a hedge fund trader where you could like really lose a lot right so in that sense you're not going to die so you can shoot for the

1:10:51Tim Ferriss:moon I do have a question I should know the answer to this but I don't so long ago this is probably 2000 2008 this is a long time ago actually I wonder if I had exposure to deep mine I invested in Founders Fund. This was a very, very long time ago. But what I did not realize internally, and I'll just read a couple of my highlights. It is absurd how many highlights I have from the Infinity Machine and all of your books. A gap opened up between Thiel and Nozick. As a general matter, Thiel doubted that going on boards was a good use of partners' time. Startups should be left to sink or swim. The art of venture capital, he liked to say, was to back contrarian ideas, not coach company founders.

1:11:27Tim Ferriss:We could spend a lot of time just on that. But I'm going to move on. Most venture partnerships decide on investments by voting. If a handful of partners see hair on the deal, the deal will be rejected. But Thiel had taken the unusual position that collective decision-making should be avoided. The way he saw things, if investments were chosen based on voting, the Founders Fund portfolio would consist of middle-of-the-road startups to which nobody objected. And then dot, dot, dot. This comes back to the power law, right? Given that all the profits and venture come from a few improbable moonshots, this sort of consensus portfolio would deliver mediocre performance.

1:11:59Tim Ferriss:So, and I'll paraphrase now, Thiel empowered the partners to go all in with their gut slash intuition. My question is, how is that governed in any way? Of course, if anyone gave 10 out of 10 conviction and then lost money consistently, they would presumably be sort of removed from the partnership or they'd lose their ability to lead with that type of gut conviction. But do you you have any idea how that was handled internally in terms of stress testing ideas, pushing people to really put their ass on the line for these types of high conviction, but certainly very much outlier investments? Do you have any idea?

1:12:41Sebastian Mallaby:Internally, Founders Fund was very torn about the deep mind investment. And I described some of this in the book where, you know, they do the first deal and that's fine. It's$2 million. But then you get to Series B and Series C and the check size gets bigger. And so the other partners are asking tougher questions and they're saying, well, wait, is there going to be a product? And Demis said to me that his attitude was, what do you mean? Is there a product? I'm talking about artificial general intelligence. It's going to make all products revolutionized or obsolete or whatever. And you want to ask me what the widget is?

1:13:18Sebastian Mallaby:Give me a break. No, it's all of the widgets. They're all going to be changed. And if you're asking me this question, you don't get what AGI means. And so Demis was very frustrated by the other partners at Founders Fund. And I think internal within Founders Fund, there was a lot of fighting between Luke, who remained enthusiastic and committed about Demis, partly because he was the guy who would go to London and meet with him and sit in the board meetings, and he would get several thousand volts of Demis enthusiasm, you know, injected into his spine at every meeting. And he would come back buzzing with excitement.

1:13:54Sebastian Mallaby:And the other Founders Fund partners who didn't have that benefit were skeptical. And so Luke would often come to Demis and say, we got your back, we got your back, we know we're going to do the next round, we're going to lead the next round. And then actually in Series C, Founders Fund at the last minute pulled out and they put money in, but they did not lead. And so the answer to your question is there was a lot of argument within Founders Fund. As the check size grew, it was harder to have that double down on your winners kind of attitude.

1:14:28Tim Ferriss:Yeah, in this case. Oh, the fish that got away. Although they did, you know, I mean, it was a fantastic multiple on their initial money. It strikes me in reading the book that I would argue that Demis made absolutely the right decision with the Google acquisition. I mean, you mentioned also in the book how he got criticized in some UK media for like, oh, it's a giant megacorporation. The US gets our prized talent cheap kind of stuff. But looking back, I mean, he seems to have anticipated the costs and compute and just raw materials that would be required to do what he was trying to do. Would you read that the same way?

1:15:10Yeah.

1:15:11Sebastian Mallaby:I mean, I often have this debate with people in London where they say exactly as you put it, you know, this was a tragedy for UK tech. Our great champion of deep tech, you know, is bought out cheaply by Google. And I say, listen, it wasn't cheap. the acquisition price might have been$650 million, which was a bit cheap. But you know how much they put in in terms of recession development funds over the next 10 years? It was approaching$10 billion, almost a billion a year. So this was not selling cheap to the Americans. This was a cunning British trick to get a billion dollars of American R &D money into London per year for the next decade.

1:15:48Sebastian Mallaby:Terrific win. And by the way, today, there are spin-outs from DeepMind in London because the talent stayed in London. And these spin-outs are raising billions of dollars to do new AI companies. So it's terrific for the London ecosystem around King's Cross, which is this sort of cool center for tech in London, where you can get the train in one direction and be in Cambridge, which has quite a lot of good startups, you know, in one hour. Or you can get the train in the other direction and be in Paris, where there's Mistral and so forth. And it's kind of very wired into different bits of Europe. So how long does it take to get from San Francisco to Mountain View, depending on the traffic?

1:16:28Sebastian Mallaby:It can be well over an hour. So I think there is a technology ecosystem which is by no means the equivalent of Silicon Valley yet, but it's certainly unrecognizably better than it was 10 or 20 years ago.

1:16:41Tim Ferriss:What do you think the UK or Europe could do? Let's focus on the UK, perhaps. could do to increase the level of innovation, early stage, startup founding, etc. Because looking back at the power law and certainly just having spent so much time in California, there's a lot that went into Silicon Valley. And there are certain things that don't get a lot of airplay. But for instance, the difficulty of enforcing non-compete agreements in California really led to this sort of round robin of talent moving and cross-pollinating like little hummingbirds of engineering talent and so on, which may not be replicable depending on where you are.

1:17:25Tim Ferriss:But what could the UK do in your mind? If you had the ear and they were like, all right, Sebastian, tell us what to do.

1:17:32Sebastian Mallaby:A couple of things. I mean, I think the mistake that people in Europe make and Britain is part of this is to believe that there's some kind of cultural magic about Silicon Valley, where whatever it is that they're drinking in the water out there makes them think that failure is a learning experience, which is kind of weird. And the Europeans say, well, we're never going to be like that. And it's impossible for us to become as entrepreneurial as Silicon Valley. And I remind people that when Fairchild Semiconductor was founded in 1957, the eight scientists who left the Shockley lab were called, get this, the traitorous eight.

1:18:10Sebastian Mallaby:Traitorous. Why? Because, you know, it was considered treachery at the time to leave one company and go to another company. There was no entrepreneurial culture in the 1950s on the West Coast in the US. The classic business book of the time was Organization Man, about people who joined one company and stayed in it for their whole life and retired with a gold watch on their 60th birthday. So you can create an entrepreneurial culture. And that is happening bit by bit in Britain and certainly in Israel. And it's happened in China. And it's not some magic which is confined to Silicon Valley. I think it's worth making that point as a first thing.

1:18:47Sebastian Mallaby:Now, there are specific policy shifts that you need to do to make an ecosystem work. And I think you put your finger on one, which is the mobility of talent is super important. You can think of a startup ecosystem as something which circulates three elements, money, people, and ideas. And you circulate those and you combine them in different ways. And each time you combine them, that's a new company. And each has a shot on goal and most of them fail. But all of a sudden, if you circulate these components fast enough, you do get product market fit. And then you get these 10x plus returns. Now, in Britain, when you raise a new round, a series B, say, and you've got nine months of runway to build to the next stage from your company.

1:19:37Sebastian Mallaby:And you identify the three key talent that you're going to bring into the company and make it happen. And then they turn around to you and say, well, I can come in six months. That's a death sentence. That's horrible. We call it gardening leave in Britain. That is an appalling idea. We've got to get rid of those gardens and we've got to let people move fast. Another thing is tech transfer out of universities. In the US, there's the Baiduil Act. There are these very sophisticated tech transfer offices which are generous to the entrepreneur in terms of not demanding too much flesh as somebody exits.

1:20:14Sebastian Mallaby:And that's essential for making the startup work. And in Europe, the attitude is, we're the university. We deserve a lot of skin in the game here. We want 50 % of the upside. Well, in that case, the startup will never happen. And I say to these Europeans, go visit Stanford. They're very generous to their entrepreneurs. They seem to be okay financially. Because if you help the entrepreneur, you know, you'll get the donations later. It's all good. And so I think those are just two things.

1:20:43Tim Ferriss:Which started a long time ago in the US. You look at the origins of Genentech and so on. I mean, it's the genesis of so many, not just companies, but industries effectively in the US. Do you think Demis would have built DeepMind if he had not read Ender's Game.

1:21:03Sebastian Mallaby:Can I just tell the Ender's Game story to begin with?

1:21:06Tim Ferriss:And also a bit of trivia for folks, I believe, and not to make this more difficult, but when Mark Zuckerberg first had a profile on Facebook, the only book listed was also Ender's Game.

1:21:21Sebastian Mallaby:Oh, I didn't know that.

1:21:22Tim Ferriss:I believe that's true. That's fascinating. So hop into it with Demis. Ender's Game.

1:21:27Sebastian Mallaby:So right at the beginning of my interviewing of Demis, we were having the second meeting, which was a dinner. And he told me to read a couple of books before we had the dinner. And one of them was Ender's Game.

1:21:39Tim Ferriss:What were the others just before you continue?

1:21:42Sebastian Mallaby:It was a book by David Deutsch called The Fabric of Reality.

1:21:48Tim Ferriss:Yeah, the light read.

1:21:49Sebastian Mallaby:Yeah. I read Ender's Game as a result, and I hadn't read it before. And as I was reading it, I was thinking to myself, okay, so this is a story about a sort of boy hero who saves the entirety of humanity from an invasion of the planet by the space aliens. Is Demis telling me that that's how he sees himself? That he's like saving all of humanity with AI? Because it'd be a bit much to believe that, but it would be even more to have the temerity to tell the guy who's writing a book about you that that's how you see yourself. Like most people wouldn't expose themselves in that way. I thought, is Demis really thinking this?

1:22:31Sebastian Mallaby:So then I go to have the dinner and he says, I hope you read Ender's Game because that's really how I see myself. And I gave the book to my wife so she could read it so she could understand me better because I really identify with Ender.

1:22:43Tim Ferriss:Yeah, it's wild. It's wild. It's a great book. I mean, I haven't read it in decades, but it is a fantastic read as I remember it.

1:22:53Sebastian Mallaby:Yeah, I mean, reading it, I must say, as a mature adult, I thought it was not that well written. But the idea of it is good. The idea is sticky. Absolutely. This image of this kid who sacrifices everything to dedicate himself to the craft of fighting the aliens and, you know, withstands ridicule and bullying from his peers and fights back. It's an appealing image, and that's what hooked Demis. But to answer your question of earlier, you know, he would have done AI anyway, because he read Ender's Game, actually, when he was already kind of around 30. And he'd had, unbelievably, the determination to build super intelligence from when he was about 17.

1:23:37Sebastian Mallaby:I mean, that is wild as well. I mean, the early conviction is just extraordinary.

1:23:42Tim Ferriss:Did he ask you to read Goodl Escher Bach, Eternal Golden Braid? I will admit to you, I think Dustin Moskovitz, a lot of technologists, very, very, very good technologists recommend this book or cite it as part of their own journey to building something incredible. I think I'm too dumb to read that book. I had so much trouble. I've had so much trouble. I've yet. I've still not finished that book. I don't know. Hey, do you have any recommendations to somebody who's maybe lacking a few IQ points because he was born on Long Island as to how to navigate that book?

1:24:20Sebastian Mallaby:I have to admit, I was told by Demis that this meant a huge amount to him, that he'd read it in his late teens. And that was when he really became convinced that he could build AI because the argument in the book is that whatever the human brain can do, computers will be able to do one day, that the human brain operates on ones and zeros. And therefore, if you could build big enough compute, you should be able to replicate the intelligence of human brains. And that was the sort of insight that got him hooked on the idea. So I went off and I tried to read it. I would say I got like 150 pages in and got bogged down.

1:24:58Sebastian Mallaby:I mean, it is a difficult, challenging read. But at least I kind of extracted the essence that meant something to my subject, to Demis.

1:25:08Tim Ferriss:You know who would be great for helping me to understand this? LLMs. Right. I'm going to give that a shot. So you've explained this to a sixth grader or explain it to a six-year-old, maybe even better. A couple of questions and we'll start to lay on the plane. If you had to write another book on a figure in the world of AI, they could be relatively unknown or they could be incredibly known. Who would that person be? Demis is off the table. I might want to take Sam off the table just to make it a little more interesting. Who would it be if Sam's off the table and Demis is, of course, off the table?

1:25:46Sebastian Mallaby:Well, I guess Dario. Yeah. I think even if you left Sam on the table, it would be Dario. I mean, I think he's just a fascinating, fascinating figure, as well as being the current leader, for the reasons I was saying earlier.

1:26:00Tim Ferriss:Of Anthropic, for people who don't recognize the name. Man, you know, I'll share it. This is not really, well, it is germane to the topic of conversation, but I'm working on a blog post right now and it's about disruption due to AI and how it's not three years in the future. It's not one year in the future. These are book sales across my entire book catalog and it's not limited to print. This is all format. Okay. So I'll give you some numbers and then I want you to tell me what happened to initiate this. Okay, 2022, stasis, pretty consistent. My book royalties are an annuity, predictable. 2023, minus 5%.

1:26:41Tim Ferriss:2024, minus 13%. 2025, minus 46%. And 2026 so far on track to be at least negative 57%. What happened at the end of 2022?

1:26:56Sebastian Mallaby:Chat GPT.

1:26:58Tim Ferriss:GPT 3.5. it's just wild. It's really, really wild. I mean, this stuff is coming fast and I really flip and flop. I feel like I waffled perhaps too much between these two. I go from the very, I would say, moderate, well-reasoned positioning of Benedict. And I agree with so many of his points to believing that all of this is just coming so much faster than anyone can even comprehend due to the sort of recursive self-improvement.

1:27:31Sebastian Mallaby:For the record, I think that it is much bigger than mobile, much bigger than internet. This is so general, cognitive capability which can span any human task. I think the niggle is simply, how long does diffusion take?

1:27:47Tim Ferriss:Yeah, right. And just to give an example of that, I invest in quite a few cloud tech companies and other sciences. And if you look at, say, AlphaFold, right, I mean, absolutely merited a Nobel Prize. We didn't mention that about demos. But it's one thing to design molecules. It's quite another to deliver it to target tissue. So like the deliverability of that, sort of a metaphor for AI in a way. It's like, okay, great. We have this pristine, perfect molecule. How do you get it to the right place. And at the same time, I'm an investor in a company called Lila, Lila Sciences, and what they're doing is producing a proprietary data set by automating wet labs using AI.

1:28:35Tim Ferriss:And I'm going to simplify it, but they have gigantic wet labs where they can run in parallel thousands of experiments that from the very first step of hypothesis generation through to the end of the scientific method is all run autonomously by AI. And I bring this particular example up because even I want to say six months ago, 12 months ago, they are producing discoveries that are really non-trivial. It's already happening now. This is not a year in the future. This is happening now. So when you flash forward to think about the potential exponential improvement, and I still, to be honest, sometimes when people talk about like exponents, exponents, humans aren't good at thinking exponentially.

1:29:24Tim Ferriss:I'm like, yes, that's true. But outside of more laws, why would AI capabilities or LLM parameters or however you want to measure it automatically improve in exponents? I don't actually quite understand that. But once we get to the sort of recursive self-improvement, it's like, okay, I can see how that starts to approach a vertical wall.

1:29:41Sebastian Mallaby:I agree with you. I think one experience from writing the book is simply that when you're close to the people inside the labs, and it wasn't just Demis, I interviewed a hundred of these AI insiders, you realize that the stuff in the pipeline is enormous. And you also, I think there's a kind of popular misconception, which is there is this thing called AI, and it kind of happened when Chatty PT came out. So now we've got it, and we're kind of getting used to it, and that's in the rearview mirror. No, no, no, no, no. This thing is changing the whole time as anybody who looks closely knows. And if you think back, the progression is wild.

1:30:17Sebastian Mallaby:You know, you get this system in end of 2022 which hallucinates non-stop. Then you plug in GPT-4 kind of six months later, whatever it was, and the hallucination radically reduces. Then it goes multimodal so it It can do video and audio. And in the meantime, it's got a very long context window. So you can plug in an entire Tolstoy novel and ask questions about it. Then it starts to do the reasoning stuff and can do logic and math. Then it becomes agentic. Then it's like coding for you. And all of these changes are packed into three and a half years. And I agree with you. I think the next three and a half years are going to be even more wild.

1:31:02Sebastian Mallaby:So I think there's a big gap between the inside and the outside view of this.

1:31:06Tim Ferriss:Yeah, that's where these comparisons to the Industrial Revolution just completely fall apart. Right. On so many levels. I have one or two remaining questions for you. The billboard question. I ask this a lot. It can be a fun one. If you could put anything on a billboard, metaphorically speaking, for millions, billions of people to see, could be anything, image, quote, question, preferably not commercial. What might it be?

1:31:33Sebastian Mallaby:A billboard which lots of people are going to see, I would put, prepare your mind. And this is a saying which is originally Louis Pasteur, I think the scientist, who said, chance favors the prepared mind. If you're ready for things, you can make the most of the opportunity that comes your way. The amazing thing about this saying is that it's come up randomly in different contexts. in different books I've done. So when I was writing about venture capital, Accel Capital, and one of the founders, Arthur Patterson, used this phrase as a description of how he wanted Accel to invest, that they would run these kind of scenario exercises where they would think, okay, there's a new technology coming down the pike.

1:32:22Sebastian Mallaby:What kind of company needs to be built to make the most of that new platform? What type of entrepreneur is going to fit this opportunity? What should we be expecting so that when the person walks into the office, into the conference room and pitches to us, we already know 90 % of what he says because we've prepared our minds. And that way we can make a good judgment and a fast judgment if it's a competitive situation. So I kind of wrote about the prepared mind in the context of venture capital. And then I'm doing the infinity machine and I'm interviewing Ilya Satskeva from OpenAI and I'm asking him, why was it you who understood the significance of the transformer architecture when it came out?

1:33:04Sebastian Mallaby:Immediately, like on the day it was up on the website, you read it, you ran down the corridor, you went to see your collaborator, Alec Radford, and you said, we're going to build a language model on top of this architecture. How did you see it so quickly?

1:33:16Tim Ferriss:Well, not only that, he said, stop everything you're doing. Right, right, right.

1:33:19Sebastian Mallaby:So you have this vision of the kind of, you know, over-caffeinated, charismatic, seizing on the engineer and saying, drop it, whatever you're doing. And, you know, in his answer was prepared mind, that he'd been thinking about how you model sequential data ever since his PhD in Canada. And when he saw the solution, this was what he'd been waiting for, for like a decade. And so he could jump on it. And then when you start thinking about prepared mind, you know, you would probably remember this better than I do, but wasn't there a Seattle Seahawks Super Bowl final against the New England Patriots, where the New England quarterback does an interception in the last second of play and clinches the victory.

1:34:01Sebastian Mallaby:And when he's asked after the play, how do you know to make that run? How do you know where the quarterback was going to throw the ball? The answer was prepared mind. Basically, he didn't use that phrase. But in training, they had studied the play that the Seattle Seahawks were going to make. And they knew that given a certain formation, when the ball was snapped back, there was a certain pass that was coming. So the guy just takes off and he runs right into where the ball comes and he catches it and intercepts and New England wins. And so that's a prepared mind in sports. And the other reason, last thing, why I would put on the billboard, prepare your mind, is that for the age of artificial intelligence, this is what we need to hear.

1:34:42Sebastian Mallaby:And this is a serious point. The risk with large language models is that we just get lazy and whenever we need to know something, we just get it to tell us what to think. That is not the route to happiness or satisfaction or anything. We need to continue to do the hard work of preparing our minds because that's what makes us people. I think, therefore I am. And so I think prepare your mind is entering a time when it becomes a more important slogan than ever.

1:35:14Tim Ferriss:How do you do that for yourself? What guardrails or policies have you established for your own use of AI? And it makes me also think of going to the gym, lifting weights, getting in cardio. You don't have to do that, but it is beneficial for you on a lot of levels. And people, some people find it quite enjoyable and hence they do that. And I'm wondering what the equivalent is for knowledge workers or people who are preparing their minds and don't want to become impotent in the way that people with directions have mostly become impotent because of Google Maps and other tools like that. So what do you do for yourself personally, or how are you thinking about that?

1:35:55Sebastian Mallaby:The first thing I think is the Google Maps analogy is the wrong one in the sense that it's fine to offload a very specific mental task, which to most people is a pain in the neck, and let the machine do that for you. It's not fine to offload all thinking. The point of offloading something should be you get to focus your mental energy more on the other stuff that you really get satisfaction and meaning from. And so for me, what that means is that I'm very happy to use large language models to learn about the scientific output of somebody I'm going to interview next week. All of these AI papers are on archive, and the model has ingested all of them.

1:36:40Sebastian Mallaby:And the model is extremely good at telling me, okay, the scientist you're seeing next week has these three papers, and the progression between the three papers is this and this and this, and the comparison with the person you saw two weeks ago is this and this and this. You know, you learn a lot from the system, like really bootstraps you to learn faster. So that's helping me to think more, not to think less. It's cutting out the time it would take me to go find all the papers by myself and then labor through them. It's cutting to the chase and nourishing me intellectually. And by the way, I'm not worried about hallucination because I'm going to interview the human scientist anyway, so I get to cross-check it all.

1:37:21Sebastian Mallaby:What I would never do is get the AI to write because frankly it's not very good at long form. In fact it really sucks. It's fine for writing an email although I don't do that either because I like writing. But it really is. I've tried it once. It's terrible for anything longer than about 800 words. But even if it could do it I don't think I would ever outsource that because that's me. This is what I do. This is the thinking process. I think through my writing. I come to understand what I understand and think what I think and believe what I believe through writing. And I'm not going to give that up.

1:37:59Tim Ferriss:I'm letting out a pensive exhale because I was thinking of this, a friend said to me, well, I'll give him credit, Kevin Rose. At one point I was, I wouldn't say complaining, observing that AI couldn't do X or it wasn't very good at Y. He said, when was the last time you tried that? And I was like six months ago. And he's like, try it again. And so So the rules will become really important as also the power of these things increases. I want to say it was the New Yorker. There was a piece in the New Yorker, it might have been the New York Times, with some very famous, I want to say novelist, could have been Pulitzer Prize winner in literature, somebody at the top.

1:38:36Tim Ferriss:And they took three or four pieces of their own writing, had AI generate three or four pieces of writing in their voice and gave it to professional readers, editors, and so on. and it wasn't clear. People couldn't figure out. They claimed that what he or she wrote was air. How long was the piece of writing? That's the, I knew that was the question you're going to ask and I don't recall. So I want to go back and look at that piece to see.

1:39:01Sebastian Mallaby:So there was a story precisely like that from an economist writer who's very funny and also does podcasts. And he ran that experiment and it was just as you said, you know, his friends who were professional economist journalists couldn't tell, which was the witty column that he'd written versus the equally witty ones, which the LLM had generated. And he was very pissed off with this. I take your point. I mean, for now I can be all complacent and say, yeah, it only works for 800 words. It doesn't work for, you know, a whole chapter, which is 20 pages long, but no doubt it'll get better and better.

1:39:38Sebastian Mallaby:But I still think I'm going to cling on to the thing that makes me, me.

1:39:42Tim Ferriss:For sure. 100%. And I think doing the thinking, preparing your mind, in part asking that question, which is not an easy question, perhaps there's a different way to phrase it, but like, what are the things that make me, me? So you don't accidentally make sacrifices that start to erode your sense of self, but also sense of self-worth, preparing your mind. Sebastian, everybody should check out The Infinity Machine. It's outstanding. The Infinity Machine, subtitled Demis Hassabis, Deep Mind, and the Quest for Superintelligence. And lest people make the wrong assumption, this is not here's the latest and greatest in AI.

1:40:26Tim Ferriss:It is the story of an incredible mind, a whole cast of kooky and fascinating characters. It's about a noble quest. It's about the pitfalls and promises of entrepreneurship. It contains so many different levels. And if you want to also have a basic understanding of what it is from the ground up that came to be colloquially referred to as AI or LLMs, this is a great book for that. It really lays out kind of the nuts and bolts and how this evolved over time in a way that I think is intelligible to non-engineers. So everybody should check out the Infinity Machine. Sebastian, is there anywhere else you would like to point people or anything else you'd like to say as we wind to a close?

1:41:12Tim Ferriss:Yeah, you stumped me on that one.

1:41:16Sebastian Mallaby:I've enjoyed the conversation. I'm happy to leave it there. Thank you for doing it, Tim. It's been great.

1:41:21Tim Ferriss:Absolutely. I'll give one more link for folks. If they want to find you on X, that's C.S. Maliby, at C.S. Maliby. And we'll link to everything we've discussed.

1:41:30Sebastian Mallaby:You just said that again. It's at S.C. Maliby, not C.S.

1:41:34Tim Ferriss:You know what? That is something that I have done since I was a little kid. It's not typically dyslexia. It's usually dysgraphia. So let's try that again. S.C. Maliby, at S.C. Maliby. Wow, it's been years since I did that. I used to do that all the time. I literally used to write my name backwards as a kid. and now I would just read it right to left as if it were Hebrew or Arabic.

1:41:56Sebastian Mallaby:And now you're a best-selling author. It's amazing.

1:41:59Tim Ferriss:Go figure. It's a miracle I got this far without LLMs earlier. Well, Sebastian, thank you so much for the time. Really enjoyed the conversation. And for people listening, we will include links to everything we've discussed, all the characters and everything else at Tim.blogs slash podcast. Just search Sebastian. I'm pretty sure that... Oh, actually, we have Sebastian Younger, So there are two Sebastians, but if you search Malaby, M-A-L-L-A-B-Y, it'll be very easy to find this. And until next time, be just a bit nicer than is necessary, a little bit kinder than is necessary to others, but also to yourself and prepare your mind.

1:42:37Tim Ferriss:Thanks for tuning in. Hey guys, this is Tim again. Just one more thing before you take off. And that is five bullet Friday. Would you enjoy getting a short email from me every Friday that provides a little fun before the weekend. Between one and a half and two million people subscribe to my free newsletter, my super short newsletter called Five Bullet Friday. Easy to sign up, easy to cancel. It is basically a half page that I send out every Friday to share the coolest things I've found or discovered or have started exploring over that week. It's kind of like my diary of cool things. It often includes articles I'm reading, books I'm reading, albums perhaps, gadgets, gizmos, all sorts of tech tricks and so on that get sent to me by my friends, including a lot of podcast guests.

1:43:22Tim Ferriss:And these strange esoteric things end up in my field, and then I test them, and then I share them with you. So if that sounds fun, again, it's very short, a little tiny bite of goodness before you head off for the weekend, something to think about. If you'd like to try it out, just go to tim.blog slash Friday, type that into your browser, tim.blog slash Friday. Drop in your email and you'll get the very next one. Thanks for listening. You guys know I love wearables. I'm sure you do as well. And they're great, but they give you data. Typically, they do not give you solutions. That's why I absolutely love The Pod by this episode's sponsor, AteSleep.

1:44:01Tim Ferriss:I've been using their stuff for many, many years now. It fits over your existing mattress tracks your heart rate with 99 accuracy plus respiratory rate hrv and sleep stages it is wild how much it correlates accurately to the stuff that you wear on you then the pods autopilot analyzes your biometrics and automatically adjusts your bed temperature while you sleep with independent temperature control for couples also important for domestic peace users report falling asleep up to 44 faster this matches with my experience i've experimented with all all sorts of stuff, countless sleep aids, and I've yet to come across a better solution that both measures and fixes my sleep within the same system.

1:44:45Tim Ferriss:Summers don't need to mean terrible sleep. So go to eightsleep.com that's spelled out E-I-G-H-T, eightsleep.com slash Tim and use code Tim for$350 off of the pod five with their 30 day trial and free returns. You can try it out risk-free. So check it out, 8sleep.com slash Tim. This episode is brought to you by AG1, which I have taken for more than a decade. Man, it's coming up on close to 15 years, I would say. Long-time listeners have heard me call it my nutritional insurance, and that is exactly how I use it. The news today is they just launched AG1 Pro, designed for people focused on performance and longevity.

1:45:25Tim Ferriss:I'll be traveling soon. I'm actually packing today and I'm throwing a bunch of these in my bag to cover my bases while I'm on the road. And I have noticed, perhaps some of you have, that it is harder and harder to put on muscle and to keep on muscle. I am a young 48 years old and I've noticed that change. So past a certain age, you may be that age. Maintaining lean muscle becomes priority number one, and AG1 Pro is formulated to support exactly that. It includes HMV, a clinically studied compound that supports protein synthesis and helps reduce muscle breakdown, plus zinc carnosine to support the gut lining.

1:46:05Tim Ferriss:So remember, training is only half the equation. You have to feed the machine. And as you get older, there's certain things that can help turn the dials just a bit in your favor. Recovery is where the work compounds on AG1 Pro is built to support it. It's NSF certified for sports. So what's on the label is what's in the pouch. Get your free AG1 Pro Yeti shaker in the welcome kit at drinkag1.com slash Tim. That's drink AG1, the number one. So D-R-I-N-K-A-G, the number one.com slash Tim.