Hours of interviews, cut down to every distinct point.

2017–2025 · changed view

“Wealth is the thing that you really want. Wealth is assets that earn while you sleep. Wealth is the factory that with the robots as cranking out things. Wealth is the computer program that's running at night that's serving other customers. Wealth is even money in the bank that is being reinvested into other assets and into other businesses. Even a house can be a form of wealth, because you can rent it out — although that's probably a lower use of productivity — but land and actually doing some commercial enterprise. So my definition of wealth is much more businesses and assets that can earn while you sleep.”

2023–2026 · changed view

“The observation is that don't die is the most played game by everyone all over the world every day. You and I breathe every few seconds, we look both ways before we cross the street, we all play don't die. So it's played more than capitalism, it's played more than any religion. It's the most played game in the world.”

2021–2025

“Freedom is freedom from the mind. The only reason that any human being is not free is because of his mind. It is the mind which creates his tortures. It is the mind which creates his anxieties. It is the mind which creates his conflicts. It is the mind which creates his rules. All that he is confined by, all that he is imprisoned by, is the mind, not anything else.”

2021–2025 · changed view

“Really these are two halves of the same coin because what crypto really does is its property and property rights and speech rights and encryption and civil liberties outside of the control of abusive states and if your state by the way is mostly convincing rather than coercing then people don't really need to use crypto quite as much I mean you can use it for lots of other things you'd use it for getting around centralized companies not just states because you you know the power of Facebook or a Twitter is also something you can decentralize with crypto which is very important so even if you don't buy the state argument all the corporate argument is important but the more abusive a state is like a Venezuela the more important crypto is it's kind of that last bulwark the check and balance is actually outside the state.”

2019–2026

“I didn't agree with that post. If you look at labor market data, the returns to IQ as it translates into wages, they're amazingly low. They're pretty insignificant. People who are very successful, they're very smart, but they're people who have say eight or nine areas where they're like, on a scale of 1 to 10, there are nine. Like they have one area where they're just like an 11 and a half on a scale of 1 to 10. And then on everything else, they're an eight to a nine and have a lot of determination. And that's what leads to incredible success. And IQ is one of those things, but it's not actually that important. It's the bundle, and the bundles are scarce. And then the bundles interacting with the rest of the world. Like just try going to a mid-tier state university and sit down with the committee designed to develop a plan for using artificial intelligence in the curriculum. And then come back to me and tell me how that went, and then we'll talk about bottlenecks.”

1995–2023 · changed view

“But of course, when people use the word common sense, what they mean is uncommon sense, because the standard human condition is ignorance and stupidity. They say old Joe has common sense, they mean as he has uncommon sense, I guess.”

2017–2026

“And I would say that is, that forms the basis of my philosophy, which is that we don't know the meaning of life, but the more we can expand the scope and scale of consciousness, digital and biological, the more we're able to understand what questions to ask about the answer that is the universe. So I have a philosophy of curiosity.”

2023–2026 · changed view

“So in the early 90s, when we first started the company, we observed that in a software program inside it, there are just a few lines of code. Maybe 10% of the code does 99% of the processing. And that 99% of the processing could be done in parallel. However, the other 90% of the code has to be done sequentially. It turns out that the proper computer, the perfect computer, is one that could do sequential processing and parallel processing, not just one or the other. That was the big observation. And we set out to build a company to solve computer problems that normal computers can't. And that's really the beginning of Nvidia.”

2022–2026 · changed view

“Oh, I think it's very different. What I meant by no one is crazy is that it is so easy for people to look around at society and how people are spending their money and saving their money and investing their money and say, why the hell would anybody do that? Why would you waste your money on this? Why would you hoard your money like that? And actually, think, if you peel back the onion layer of what's going on in those people's lives, no one is really that crazy with how they spend their money or save their money. It makes sense to them in that moment. My brother-in-law is a social worker. He works with very disadvantaged kids, kids who are abused at home, who don't have homes. And a lot of those kids, not surprisingly, do very poorly at school. They misbehave, they get in fights, they don't go to school. And it is so common for those kids that the teacher will look at those kids and say, why are you acting like this? Why don't you do the right thing? It's so obvious. And he says there's a phrase in social work: all behavior makes sense with enough information. That if you look at what those kids are dealing with in life, it would all make sense to you why they are misbehaving at school. And I think that's a powerful idea for a lot of things in life, that all behavior makes sense with enough information.”

2023–2026 · changed view

“I love that you asked them together, because I think they're super related. I think the economic transformation will drive much of the political transformation here, not the other way around. My working model for the last five years has been that the two dominant changes will be that the cost of intelligence and the cost of energy are going, over the next couple of decades, to dramatically, dramatically fall from where they are today. And the impact of that, and you're already seeing it with the way you now have like, you know, programming ability beyond what you had as an individual before, is society gets much, much richer, much wealthier, in ways that are probably hard to imagine. I think every time that's happened before, it has been that economic impact has had positive political impact as well. And I think it does go the other way too, like the socio-political values of the Enlightenment enabled the long-running technological revolution and scientific discovery process we've had for the past centuries. But I think we're just going to see more. I'm sure the shape will change, but I think it's just long and beautiful exponential curve.”

1998–2020

“So, what we really want to do is buy businesses that we would be happy to own forever. It's the same way I feel about people buy Berkshire. I want people to buy Berkshire to plan to hold it forever. They may not for one reason or another, but I want them at the time they buy it to think they are buying a business that they're going to own forever. And I don't say that's the only way to buy things. It's just that that's the group I want to have join me because I don't want to have a changing group over time. I measure Berkshire by how little activity there is in it. If I had a church and I was the preacher and half the congregation left every Sunday, I wouldn't say, 'Oh, this is marvelous to have all this liquidity among my members, you know.' This terrific turnover, you know. I would rather get a church where all the seats are filled, you know, every Sunday by the same people. Well, that's the way we look at the businesses we buy. We want to buy something that we're really happy to own virtually forever. And we can't find a lot of those.”

2023–2026 · changed view

“So the main thesis on the essay is that what we're dealing with here is intelligence, and it's really important to kind of talk about the sort of very nature of what intelligence is. And fortunately we have a predecessor to machine intelligence, which is human intelligence, and we've got, you know, observations and theories over thousands of years for what intelligence is in the hands of humans. And what intelligence is, right, I mean what it literally is, is the way to, you know, capture, process, analyze, synthesize information, solve problems. But the observation of intelligence in human hands is that intelligence quite literally makes everything better. And what I mean by that is every kind of outcome of like human quality of life, whether it's education outcomes or success of your children or career success or health or lifetime satisfaction, by the way, propensity to peacefulness as opposed to violence, propensity for open-mindedness versus bigotry, those are all associated with higher levels of intelligence. Smarter people have better outcomes than almost as you, right, in almost every domain of activity.”

2013–2020 · changed view

“Hammurabi's rule, and it says the following: if the architect builds a building and the building collapses and kills the owner of the building, what should happen to the architect? It should be put to death. I mean, it was Babylon 3,800 years ago, so they were not, you know -- didn't have a very sophisticated legal code, all right. And then of course if it kills the firstborn, the firstborn son of the architect is put to death. So they're really looking for symmetry. Now what was the idea of Hammurabi's code? It was not eye for eye -- eye for eye limiting damages, you know, and making things balanced. No, it was that you cannot walk away from risks you've created for others. You should own your risks. That's the idea of Hammurabi's code.”

2015–2025

“One sort of question is always like, what great business is nobody building? There's a more intellectual version of it, which I always like to ask as an interview question. It's: tell me something that's true that almost nobody agrees with you on. And this turns out to be a shockingly hard question for people to answer, and it's hard because, for one thing, it's hard because we've just been taught that truth is conventional. The truth is whatever everybody agrees on, and so coming up with something new that people don't already agree on puts us in a very uncomfortable sort of a place. It sounds like it's really hard to do this, and so you have to be really, really brilliant, so we think. But the other reason it's a very tough question to answer is because you sort of think about the literal dynamic of an interview, where I'm asking the question to somebody, and then give me an answer, and it's like, oh yeah, I already thought that, or I agree with that — that's the wrong answer. And so that the right answer is one that, in some sense, people might not agree with, and so it, I think, often requires quite a bit of courage to come up with an answer that people haven't thought of.”

2022–2026 · changed view

“Neural networks have done language modeling for a long time. So really what's new or interesting or exciting is just realizing that when you scale it up with a powerful enough neural net, Transformer, you have all these emergent properties, where basically what happens is if you have a large enough data set of text, you are in the task of predicting the next word, you are multitasking a huge amount of different kinds of problems. You are multitasking understanding of, you know, chemistry, physics, human nature. Lots of things are sort of clustered in that objective. It's a very simple objective, but actually you have to understand a lot about the world to make that prediction.”

2015–2026 · changed view

“Something we learned from YC was the importance of focusing on very concrete, easy to explain customer problems. Like it's very easy to hallucinate or to imagine some customer problem that's not actually something viscerally felt by a person who would pay money. And so over the course of, in part working on Octomattic together We encountered this issue of it being really annoying to deal with the movement of money or payments, whatever, on the Internet. And on the one hand, it seemed like an obviously good idea in the sense that nobody liked the existing ways of doing so, and they were broadly extremely unpopular and kind of antiquated and legacy. And yet to, like, fill out this paperwork and go to the bank in person and the paperwork was in Latin and just like, it was all bad. But then the flip side is it just seems kind of ridiculous the two kids would start a financial services business and Fintech didn't exist as a sector at the time. Like, the word literally didn't exist. And so it's just kind of, you know, we felt like the proverbial squirrels in a trench coat trying to masquerade as a sort of a real business or as serious adults, but obviously knowing nothing coming in about the space. And certainly a lot of people we met and pitched or banks or partners or whatever that we talked to, I mean, they didn't literally laugh us out of the room, but you kind of see them looking for the button to call security under the desk and have them haul us out, because it just seems so improbable. So anyway, I think it both seemed like an obviously good idea in that people. People really wanted this, but also a bad idea that nobody took it seriously. But I think the fact that it was ultimately, the fact that it was grounded in such a concrete, actual real user problem saved us.”

2013–2026 · changed view

“But because we borrow, we have cycles. This isn't due to any laws or regulation, it's due to human nature and the way that credit works. Think of borrowing as simply a way of pulling spending forward. In order to buy something you can't afford, you need to spend more than you make. To do this, you essentially need to borrow from your future self. In doing so you create a time in the future that you need to spend less than you make in order to pay it back. It very quickly resembles a cycle. Basically, anytime you borrow you create a cycle. This is as true for an individual as it is for the economy. This is why understanding credit is so important because it sets into motion a mechanical, predictable series of events that will happen in the future.”

2009–2025 · changed view

“About a month after we started Y Combinator, we sort of stumbled upon the phrase that became our motto: make something people want. And we've changed our minds about a lot of things since then, but if I were choosing again today, this is still the one I would choose. I don't think you can get it down to three words — of course, if you could, I would be the first to try. Another thing we tell startups, probably the second most common thing we tell startups, is not to worry too much about the business model at the beginning. Don't worry too much about money — not because it's not important, but because it's so much easier than making something great.”

2023–2026

“If you're under 30, you have no reason not to go 100% all-in on your goals. The downside is nothing and the upside is everything.”

2022–2026

“You want this peak in cortisol as high as possible in the first 60 minutes of your day. Everyone thinks you want low cortisol, but here's the deal. High cortisol in the early part of your day is what provides for elevated mood, focus, attention, your ability to learn. It's also what establishes low cortisol in the afternoon and at night which is a requirement for falling and staying asleep.”

2021–2026

“Optimism as I define it has to do with knowledge. It's not a prediction of the future, it's an explanation of failure. If you explain failure as being inevitable, or due to some malevolent force, insuperable malevolent force, or just the way things are, then that's a recipe for stasis, which is a recipe for failure and eventually death. Therefore I think that all failure has to be explained in the form: the reason we didn't succeed is that we didn't know how to, and the knowledge of how to is in principle attainable. We don't have it now, but we could have it in the future if we do the right thing. In fact, it follows from this whole conception of knowledge that we've been talking about all this time that optimism has to be true. Otherwise there would be a limitation, which would mean the supernatural and all that stuff that — the arguments are watertight all the way from the scientific worldview to optimism, in my view.”

2015–2025 · changed view

“The ultimate foundation are mathematics and logic. If you understand logic and mathematics, then you have the basis for understanding scientific method. Once you understand the scientific method, then you can understand how to separate truth from falsehood in other fields and other things that you're reading. So be very careful about reading other people's opinions, and even be careful about reading facts, because so-called facts are often just opinions, but, you know, with a veneer around them.”

2023–2026 · changed view

“And even now, if you give five anti-aging experts the same scientific papers and ask them to develop a protocol for you, you'll get a different protocol from every single one. They're not going to agree. There's no way to go out there and get consensus in the world, so you need to pick a path and then measure. And I've done exactly that. So I've basically tried to punch through all the noise and say, is there actually something I can do which has some believability? That's what I've done. So I've published all my data, and so Blueprint provides people a starter, a starting point, to say I'm going to do something that I can see works, I'm going to measure myself, then iterate and improve upon it. So health and wellness is all like a religion, where the King James version the Bible supports 100 different denominations. They all say their god's one and true only. Same with health and wellness: everyone claims they're God's true health and wellness program. And I've tried to punch the whole thing to say it doesn't matter what guru status is, share the data.”

2021–2025

“Prescriptions are how-tos, hacks, five-step plans, methods, techniques. These are the only currency that exists in the world today, from the domains of the most spiritual and so-called sacred to the domains of the outright materialistic and worldly. It's literally the only thing that exists. There is no such thing as a domain or an expert or a discipline or a teaching or a book or anything that exists anywhere on this planet that is not a prescription, whose philosophy is not founded entirely and solely and exclusively upon prescriptions. Everyone has been conditioned into — everyone has been brainwashed, and the brainwashing is so complete that even the so-called wise men fall for it.”

2019–2025 · changed view

“Are new countries even possible? We know that we can found new companies like Google out of a garage, and new communities — the three billion person community that is Facebook started out of a dorm room. We know we can start new currencies like Bitcoin from a white paper. But can we start new countries similarly? That's the next step. That's what taking the internet seriously means.”

2024–2026 · changed view

“Yeah, that's what I think it will be like again. Great stagnation is over. We'll go back to those earlier rates of change, transform a lot of the world, mostly a big positive. A lot of chaos disrupted institutions along the way. That's my prediction.”

1995–2020 · changed view

“Cicero is famous for saying a man who doesn't know what happened before he's born goes through life like a child. ... If you generalize Cicero, as I think one should, all these other things that you should know in addition to history — and those other things are the big ideas in all the other disciplines. And it doesn't help you just to know them enough so you could prattle them back on an exam and get an A. You have to learn these things in such a way that they're in a mental latticework in your head and you automatically use them for the rest of your life.”

2017–2026 · changed view

“I think it's important to have a future that is inspiring and appealing. I just think there have to be reasons that you get up in the morning and you want to live. Like, why do you want to live? What's the point? What inspires you? What do you love about the future? And if we're not out there, if the future does not include being out there among the stars and being a multiplanet species, I find that it's incredibly depressing if that's not the future that we're going to have.”

2023–2026 · changed view

“And so the question is the what I'm leading up to is ultimately what is the purpose? What is the purpose of the lawyer? And has the purpose changed? What is the purpose? You know, one of the examples that I gave is that I would give is for example if my car became self-driving will all chauffeers be out of jobs? The answer probably is not because for some per for some chauffeers they for some people who are driving you they could be protectors some people they're part of the experience part of the service so when you get there they you know they could take care of things for you and so for a lot of different reasons not all chauffeers would lose their jobs some chauffeers would lose their jobs and many chauffeers would change their jobs and the type of applications of autonomous vehicles will probably increase you know the usage of the technology within find new homes and so I think you have to go back to what is the purpose of a job you know like for example if AI comes along I actually don't believe I'm going to lose my job because my purpose isn't to I have to look at a lot of documents I study a lot of emails I look at a bunch of diagrams you know the question is what is the job and the purpose of somebody probably hasn't changed a lawyer for example help people that probably hasn't changed studying legal documents generating documents it's part of the job not the job.”

2022–2026

“And the wider the channel is, the more that you can endure. And so much about compound interest, not just with money, but with relationships and careers and friendships, comes down to what can you endure? How many unknowns and the depths of those unknowns can you endure and survive? And so much, I wrote this in Psychology of Money. If you have to like sum up doing well financially in one word, I think it's survival. That's true in your career. That's true for savings. That's true for investing. Absolutely. It's just survive. It's just what can you endure? What can you put up with? And when you view it like that, you realize that like every dollar that you save is a claim check of independence that will serve you not just in the future but today.”

2023–2026 · changed view

“So I want to be very clear. I do not think we have yet discovered a way to align a super powerful system. We have something that works for our current scale called RLHF, and we can talk a lot about the benefits of that and the utility it provides. It's not just an alignment, maybe it's not even mostly an alignment capability. It helps make a better system, a more usable system. And this is actually something that I don't think people outside the field understand enough. It's easy to talk about alignment and capability as orthogonal vectors. They're very close. Better alignment techniques lead to better capabilities, and vice versa. There's cases that are different, and they're important cases, but on the whole, I think things that you could say like RLHF or interpretability that sound like alignment issues also help you make much more capable models, and the division is just much fuzzier than people think. And so in some sense, the work we do to make GPT-4 safer and more aligned looks very similar to all the other work we do of solving the research and engineering problems associated with creating useful and powerful models.”

1998–2026

“People behave very peculiarly and in terms of the reactions because they, they're human beings and they get excited when others get excited, they get greedy when others get greedy, they get fearful one other get fearful, and they'll continue to do so. And you will see, you know, if, you will see things you won't believe in your lifetime and securities markets, and the country will do very well over time, but you will see these huge waves. And then, and if you can stay objective throughout that, you can detach yourself temperamentally from the crowd, you get very rich and you won't have to be very bright. I mean, it, I'm sure you are, but that one, you know, it just, it doesn't take brains, it takes temperament.”

2023–2026

“So look, the only way civilization gets moved forward is when people like this do something new. Because civilization as a whole does not do new things. Groups of people do not do new things. These things don't happen automatically. By default nothing changes. The only way civilizational change on any of these axes ever happens is because one of these people stands up and says, no, I'm going to do something different than what everybody else has ever done before. So, this is progress, like, this is actually how it happens. Sometimes they get lionized or awarded. Sometimes they get crucified. Sometimes the crucifixion is literal. Sometimes it's just symbolic. But they are those kinds of people, and then martyrs when they go down in flames and again, this is where it really screws the people's moral judgments because everybody wants to have the sort of super clear story of like, okay, he did a bad thing and he was punished. And I'm like, no, he was the kind of person who was going to do great things and also was going to take on a level of risk and take on a level of sort of extreme behavior such that he was going to expose himself to flying too close to the sun, wings melt and crash to ground. But it's a package deal. The reason you have the Picasso's and the Beethovens and all these people is because they're willing to take these extreme level of risks. They are that creative and original, not just in their art or their business, but in everything else that they do that they will set themselves up to be able to fail psychologically.”

2008–2024 · changed view

“Problem of most theorists, problem for universities, is you go from theory to practice, the degrees of freedom from theories of practice are considerably narrow, small. The reverse is monstrous. So let me give you this for: if I leave in a small piece of ice cube on the floor, okay, you can easily get someone in second year physics students to write the equation to tell you what's going, hard to predict, how that ice cube will melt. Okay, it's a very simple thing, the second rate student could do it. All right, right, know someone from Boston that is can do it. Okay, so the, yeah, so you can predict. So, but now, conditional on seeing water on the floor, is it easy to reverse-engineer the ice cube? No. You have an infinity of ice cubes.”

2019–2026 · changed view

“And then in some ways I believe it's slowed down over the last 50 or so years Maybe 1970 or so is a is an inflection point one could one could cite it doesn't mean it's stopped altogether you know one way I've often summarized it is that we've continued to have progress in the world of bits you know computers software internet mobile internet you know maybe crypto now AI but there's been much less progress in the world of atoms.”

2022–2026 · changed view

“Yeah, so I had a blog post on Software 2.0, I think several years ago now. And the reason I wrote that post is because I kept— I kind of saw something remarkable happening in like software development and how a lot of code was being transitioned to be written not in sort of like C++ and so on, but it's written in the weights of a neural net. Basically just saying that neural nets are taking over software, the realm of software, and taking more and more tasks. And at the time I think not many people understood this deeply enough, that this is a big deal, it's a big transition. Neural networks were seen as one of multiple classification algorithms you might use for your data set problem on Kaggle. Like, this is not that; this is a change in how we program computers. And I saw neural nets as, this is going to take over, the way we program computers is going to change, is not going to be people writing a software in C++ or something like that and directly programming the software. It's going to be accumulating training sets and data sets and crafting these objectives by which we train these neural nets, and at some point there's going to be a compilation process from the data sets and the objective and the architecture specification into the binary, which is really just the neural net, you know, weights and the forward pass of the neural net, and then you can deploy that binary.”

2019–2026 · changed view

“Since I was born for example global life expectancy has increased by about six years and infant mortality has fallen by more than 50% again I mentioned the statistic you know the number of people who have left extreme poverty this is incredibly important right and so I think there's a you know we're not the first people to say it but there is a moral imperative to this kind of progress and we shouldn't lose sight of that fact.”

2018–2026

“Almost the only way that you understand what truth is, is through experimentation, and so when you go out and experiment you're going to learn a lot more about what truth is. But the essence of what I'm saying is that when you take a look at that, use truth, that find out what the realities are as a foundation, do the independent thinking, do the experimentation to find out what's true, and change and keep going after that.”

2014–2026 · changed view

“It turns out it's not that important to be smart; it's much, much more important to be determined. If you imagine this hypothetical person who's like a hundred out of a hundred for smart and a hundred out of a hundred for determination, and then you start taking away determination, it doesn't take very long before you have this sort of ineffectual but brilliant person. Whereas if you take someone who's super, super determined and you start taking away smartness bit by bit, eventually you get to some guy who owns a lot of taxi medallions — but he's still rich. Or like a trash hauling business or something like that.”

2023–2025

“I think that they have to have the power to influence, and that is across lots of things, just they have to be able to move other people, and you can define sales as the ability to get people to comply with your request, you can define leadership the same way, management, marketing to a degree is getting people to comply with a larger request, you know, publicly. But I think that fundamentally is a skill that people have to have if they're going to be successful at entrepreneurship. They have to have tremendous drive, whether that's a combination of towards or away, so they have a big mission that they really want to achieve or they have some very big fear that they're running away from. Either way I think the fuel works. Just from a pure entrepreneurship perspective, third piece is impulse control, is that they have to be able to say no to things on a regular basis for an extended period of time. And I think they have to be able to boil down the success of their business into inputs and outputs. Like, if you do not know the inputs that are going to get the output that you want, then what are you doing?”

2020–2026 · changed view

“how much dopamine is in our system at any one time compared to how much dopamine was in our system a few minutes ago, that dictates your so-called quality of life and your desire to pursue things. This is important. Your experience of life and your level of motivation and drive depends on how much dopamine you have relative to your recent experience. This is why when you repeatedly engage in something that you enjoy, your threshold for enjoyment goes up and up and up.”

2021–2026

“That's right. Knowledge isn't just a matter of belief, and it's not just a matter of human psychology. It is information that has causal power.”

2015–2025 · changed view

“Happiness is that state when nothing is missing. And when nothing is missing, your mind shuts down, and your mind stops running into the future or running into the past, they regret something or to plan something. And then in that absence, for a moment, you have internal silence, and when you have internal silence, then you are content and you're happy.”

2023–2026 · changed view

“I do think that humanity is at this critical juncture. Like we could be evolving to something spectacular. We could also get ourselves into a mess in so many ways. And I see the situation for what it is. And it's hard for me to spend time on anything other than that. But that's not where people's minds are at. People want to put on muscle and they want to lose weight and they want to not feel anxious and they want to not be depressed and they want to make more money at their job and they want more status in life and they want a partner in life. These basic things that humans want.”

2021–2025

“Rarity moves me, you know? So everything that I do, that's what I mean by how we behave, what our actions are, what our behaviors are, are all a function of what we desire most. And if you desire money most, you will price things accordingly. And you will go out and try to convince, and you will create mass marketing campaigns, and you will — and none of which are wrong. And I really honestly mean that. None of that is wrong. I'm not going to bash someone for their desire, whatever their desire is. 'Since I want to make $5 billion, cuz I really get a kick out of having that much money, and I want to have the—' That's fine. Like, I have no — I'm making the statement that whatever you desire is how you will behave. Whatever you are most driven by and moved by will be what is reflected in what you do. It'll all be colored by that.”

2021–2026 · changed view

“By analogy we have different categories for cash currency stocks bonds mortgages art et cetera all these scarce assets today the fundamental innovation of the blockchain was the ability to represent scarcity digitally because before blockchains you could you could transfer or rather you could you could copy rather than transfer something right I can copy a file to your hard drive what I couldn't do is transfer it in such a way that you knew I no longer had possession of it.”

2019–2026 · changed view

“Absolutely. Yeah, and here that is either acceptable or even admired, or to be a loner. And since so many people are outsiders and that we're all immigrants, is selecting for people who left something behind, we're willing to leave behind their families, we're willing to undergo a certain brutality of switch in their lives, makes us a nation of weirdos. And weirdos are creative. Yeah, and Denmark is not a nation of weirdos. It's a wonderful place, you know, great for them. Ideally you want part of the world to be fully weirdos and innovating, and the other part of the world to be a little kind of chicken, risk-averse, and enjoy the benefits of the innovation, and to give people these smooth lives and six weeks off and free ride. And everyone's like, oh, American way versus European way, but basically they're complements.”

2008–2023 · changed view

“If you get right down to it, what I am is a lover of the progress of civilization. That turns me on.”

2017–2025

“Well, I think it's helpful. The tools of physics are very powerful and can be applied, I think to almost any, really any arena in life. It's really just critical thinking. For something important you need to reason from first principles and think about things in the limit one direction or the other.”

2023–2026 · changed view

“When Nvidia started in 1993, we were trying to create this new computing approach. The question is what's the killer app? And the problem we wanted to the company wanted to create a new type of computing pro a computing architecture a computing a new type of computer that can solve problems that normal computers can't solve. Well, the applications that existed in the industry in 1993 are applications that normal computers can solve because if the normal computers can't solve them, why would the application exist? And so, we had a mission statement for a company that has no chance of success.”

2022–2026 · changed view

“What I think is true is that on average, people who have more money tend to have fewer bad days, but I don't know if they have more better days. I think it is more likely that when you have more money, you are less likely to wake up and be like, man, things are not going well. But that doesn't necessarily mean that you're going to wake up grinning ear to ear. Now, that's a lifestyle improvement. If you have fewer bad days, that's like great. Your life is better. That's obviously better than the alternative, but it's not happiness. It's a very different thing. And I think it can throw some people off when they aspire to happiness or they think that what they're achieving in the money they're making should make them happy. Sometimes it does. Very often it doesn't. And they wake up and they're like, 'What happened?' Like I thought this was going to make me happy, but I'm not. So, I think there's a level of expectation setting where it's like I think money can be more like a vaccine where like it can prevent a lot of misery, which is great. Vaccines are wonderful. Like we don't have polio. It's a great thing, but maybe that's a good analogy because you and I don't wake up in the morning being like, 'Oh, I'm so glad I don't have polio. So grateful for the fact we don't do that.' We don't think about it. And I think that's a lot of what money can do. It is a lifestyle improvement, but don't think it's going to make you waking up ear to ear, grinning ear to ear.”

2023–2026 · changed view

“People talk about like how many jobs is AI gonna do in five years and the framework that people have is what percentage of current jobs are just gonna be totally replaced by some AI doing the job? The way I think about it is not what percent of jobs AI will do, but what percent of tasks will AI do and over what time horizon. So if you think of all of the like-five second tasks in the economy, five-minute tasks, the five-hour tasks, maybe even the five-day tasks, how many of those can AI do? And I think that's a way more interesting, impactful, important question than how many jobs AI can do because it is a tool that will work at increasing levels of sophistication and over longer and longer time horizons for more and more tasks and let people operate at a higher level of abstraction. So maybe people are way more efficient at the job they do. And at some point, that's not just a quantitative change, but it's a qualitative one too about the kinds of problems you can keep in your head. I think that for videos on YouTube, it'll be the same. Many videos, maybe most of them, will use AI tools in the production, but they'll still be fundamentally driven by a person thinking about it, putting it together, doing parts of it, sort of directing it and running it.”

1998–2020

“I've never been pessimistic about the United States. I bought my first stock in the spring of 1942. I was 11. And if you remember, you would, you don't remember, no one watching this will remember, but we were losing the war at that time and we were getting totally creamed in the South Pacific, and the Philippines fell and the death march at Bataan and all that sort of thing. It was not until the Battle of Midway, which was a little later, the things started turning around, but I was optimistic on the country then and I've been optimistic on the country ever since”

2023–2026

“So the ultimate definition of an elite is who can get who fired, right. That's the ultimate test. Who can get who fired, boycotted, blacklisted, ostracized. Like when push, prosecuted, jailed, like when push comes to shove. I think that's always the question, who can destroy whose career? And of course, you'll notice that that is heavily asymmetric when these fights play out. Like, it's very clear which side can get the other side fired and which side can't. And so, yeah, so, look, I think we live in a period of time where the elites have gotten to be extreme in a number of dimensions. I think it's characterized by, for sure, extreme groupthink, extreme sanctimony, extreme moral, I would say dudgeon, this weird sort of modern puritanism, and then an extreme sort of morality of punishment and terror against their perceived enemies.”

2008–2024

“The whole idea of the Black Swan, everything I've been telling everyone, every person I meet, is to bet on disruption, on unforeseen, to not be harmed by silent risk. Okay, that's the idea. The first thing is don't be harmed by it.”

2015–2024

“I think that, um, all — if you're a founder or entrepreneur, what you want to aim for is monopoly. You want to aim to build a company that is one of a kind, uh, and that it's so far, um, differentiated from the competition that it's not even competing. Um, and, um, and I think this is the — conventional wisdom is always that capitalism and competition are somehow synonyms. I believe they're antonyms. A capitalist is someone who's in the business of accumulating capital. A world of perfect compet— competition is a world where all the profits are competed away. If you want to compete like crazy, um, then you should just open a restaurant in Chicago. Um, and, um, and, uh, and I think the, the great companies, like, uh, Google, sort of the paradigm example I use, has had no serious competition in search since 2002, when it definitively distanced itself from Yahoo and Microsoft, and as a result it's been generating enormous cash flows for the last 12 years.”

2022–2026 · changed view

“One more thing that I would say is that many times, when people think about education, they think more about what I would say is a softer component of diffusing knowledge. I have something very hard and technical in mind. In my mind, education is the very difficult technical process of building ramps to knowledge. In my mind, nanochat is a ramp to knowledge because it's very simple. It's the super simplified full-stack thing. If you give this artifact to someone and they look through it, they're learning a ton of stuff. It's giving you a lot of what I call eurekas per second, which is understanding per second. That's what I want, lots of eurekas per second. So to me, this is a technical problem of how do we build these ramps to knowledge. So I almost think of Eureka as maybe not that different from some of the frontier labs or some of the work that's going on there.”

2019–2024

“Part of our hope you know it's not to kind of promote any specific solution any specific I don't know aspect of it but rather that you know even though this is not what's kind of focally central in the headlines today it should be and as we think of what the world is going to look like in 50 years or 100 years it plausibly more than anything else is going to determine the shape of that.”

2018–2026

“Investing is an exchange of a lump sum payment for a future cash flow, right? So it's maybe the cash you have or even the cash you can borrow, but you can take a certain amount of cash and you buy something and what you're getting is a cash flow.”

2012–2025 · changed view

“Imagine he encounters this very beautiful attractive woman and he's like, and she says to him, here's the thing: if you ever got me, you wouldn't have a clue what to do with me. And that's what these really big ideas say to you. This phenomenon is one of the most important things you can understand about startups. You'd think really big startup ideas would be attractive, but actually they tend to repel you, and so you need to learn how to tack into that wind, essentially.”

2023–2025

“So, I routinely get asked the secret to success. And it just comes down to this. Number one, get better. Number two, never stop. If you do only those two things, you will win on a long enough time horizon. The problem is people convince themselves they no longer want something once they see the experience of how hard it really is. So I want to set this expectation for you as you head off to practice scripts, mark your calendar, and set your alarms. The work begins when your motivation ends. Just win.”

2020–2026

“What is cortisol? We I should have mentioned this earlier. What is cortisol? Is it a stress hormone? No. It's a hormone that's released during stress. But cortisol's job is not to deal with stress. Cortisol's job is to mobilize energy in the body. When cortisol goes up, glucose is mobilized and other fuels, too.”

2021–2026

“And my rather heterodox answer to that is that it didn't evolve for the purpose for which we now use it. It evolved for something else, namely for the transmission, or better put, for the reception of cultural knowledge, memes, cultural knowledge. So, many animals have cultural knowledge, but it's very hard to transmit. It involves things like mirror neurons or something where if you do this, then the ape can also do that. But if you do a purposeful thing, then as soon as you get to any kind of complex purpose, apes can't copy it.”

2015–2025 · changed view

“There's a thing called Agrippa's trilemma, and Agrippa's trilemma says that any questioning like this, why, will always end in one of three places, okay? First is infinite regress, right? Why? Because of this. Why? That. What? And just keep going forever. The second is circular reasoning. Well, A, why A? Because of B. Or why B? Because of A, right? You get trapped in that. Or the third is an axiom, and the most popular axiom is God. But it could be anything. Because of math, because of science, because the Big Bang, because the simulation, right? These are all axioms. These are all just stopping points. Saying simulation, we're in a simulation, or saying it's the Big Bang, is just another way of saying God. It's just God's a dirty word, so we don't use it as much anymore, but same thing. So you end up in one of these three dead ends, essentially, right? So there is no answer. The real answer is because, right? What is the meaning of life? Yeah, you get to make up your own answers. The beauty— if there was a single answer, we would not be free. We would be trapped, because then we would all have to live to that answer. Then we'd be bored, like robots, each one competing with each other to fulfill that single meaning more than the others. Back to signaling, like, I'm better at it than you are. But luckily there is no answer, so you just do whatever you want.”

2023–2026 · changed view

“Yeah, yeah, sleep is the most powerful drug that anyone can take. The next is probably exercise, and the next is diet. Those three things in order.”

2021–2025

“But that's how all things are created. That's how all things are created. So if that's how all things are created, then the only place to go is truth. If rain is your goal, then the only place to go is to line up with the atmospheric pressure and the moisture and the whatever parameters that make the rain fall. So truth is the only thing that isn't spiritual, and there's no abstractness. Rain isn't spiritual, right? It isn't like holy rain, it's rain. Okay? So that's how all things work. And I realized that, so for me the only path became truth. If I'm on that side, if the more I can learn about how things happen all by themselves, why wouldn't I learn that? Because that's — the outcome's only going to be listening to that. The outcome isn't going to be listening to the people. So why would I listen to those people whom the outcome doesn't listen to? There is nothing spiritual and abstract and philosophical and woo woo about what I just said. This is second grade.”

2025–2026 · changed view

“So because when they talk about AGI, they talk about AGI as implicitly a unitary thing. we will get to AGI and then it will go to infinity and we'll be like you know raptured singularity kind of thing even though that's implicit and you know the way I because I've been thinking so much about crypto and other kinds of things I was like well there's a different sort of implicit school of thought which is polytheistic agi right do we have rather than the vengeful god do we have war of the gods do we have many superhuman intelligences all from different cultural backgrounds that have the mores and the values imprinted on them. And very early on, I had I had a tweet that said at a minimum there's going to be American AI and Chinese AI. And if we're lucky, there'll be decentralized open-source, you know, crypto style AI because at the time the American AI was woke, highly woke. This is 2022. And I knew I knew that China was going to like stop at nothing to copy it. So I knew we were going to get at least two.”

2019–2026 · changed view

“I don't think economics will ever be very predictive. It's most useful for helping you ask better questions. You look at something like game theory: well, game theory never predicted USA and USSR would have a war, would not have a war, but trying to think through the logic of strategic conflict, if you know game theory, it's just a much more interesting discussion.”

2007–2023 · changed view

“Well I think it's very useful to get the capitalist perspective, namely how do you make this thing work for the people that own it, because the reality will make more sense when you go at it with that perspective. So even if you don't intend to get rich, you want that perspective, because you'll understand the reality you're dealing with better. So I think you mentally want to make yourself an investor as part of the proper training of your mind. And since you have to do it anyway to improve your mind, you may as well get rich while you're doing it. And so if you underspend your income while you're life and keep learning and collect all the bonehead experience mistakes other people made and avoid them and so on, I would confidently predict that you'll do just fine.”

2020–2026 · changed view

“If an AI cannot figure out new physics, it's clearly not equal to humans, let alone nor has surpassed humans, 'cause humans have figured out new physics.”

2023–2026 · changed view

“I think I know what consciousness is the sense of experience the ability to know self versus the ability to be able to reflect know our own self the sense of ego I think all of all of those human experiences probably is what consciousness is but why it exists versus the concept of knowledge and intelligence which is what AI is defined by today it has knowledge it has intelligence artificial intelligence we don't call it artificial consciousness artificial intelligence the ability to perceive believe, recognize, understand, plan, perform tasks. Those things are foundations of intelligence to know things, knowledge. I don't, it's clearly different than consciousness.”

2023–2026

“The definitions are my own, so I'm just making this up, but I think rich is when you have enough money to make your mortgage payment, make your car payment, you can pay off your credit card bill every month. Like, you can afford the things that you're buying, technically. Wealthy, I think, is when you have a degree of independence and autonomy. The weird thing here is that wealth is the money that you don't spend. That's what wealth is. Like, the homes you didn't buy and the car you didn't buy. It's money that you saved and invested that is going to give you independence. And that's a hard thing for people, I think, to wrap their head around, that wealth is what you don't see. Because I can see your house, I can see your car, I can see your clothes, but I have no idea what your net worth is. I can't see your brokerage account. I can't see your bank account. So wealth is always hidden, and it throws a lot of people for a loop. Because if I was looking for a role model of physical fitness, well, I can see your fitness. I can see your weight and your muscle tone, what not. It's all visible. But when you're looking for a financial role model, who do you look up to? And a lot of people, particularly young people, will look up to the guy in the mansion with the Ferrari. But that guy, for all you know, is living paycheck to paycheck. A lot of those people are. And the person who is actually wealthy and independent might be the person in the modest house driving the modest car, that you would actually want to be. If you want to be wealthy instead of just rich, you want to be independent instead of just making your monthly payments, the people that you actually want to look up to are some of the hardest people to identify in society.”

2023–2026 · changed view

“First of all, you can't ask it anything. That's very nice of you to say, but there's a lot of things that it's still embarrassingly bad at. But even if we fixed those, which hopefully we will, it doesn't continuously learn and improve. It can't go get better at something that it's currently weak at. It can't go discover new science and update its understanding and do that. And it also kind of can't — even if we lower the bar — it can't just sort of do any knowledge work you could do in front of a computer. Even without the sort of ability to get better at something it doesn't know yet, I might accept that as a definition of AGI. But the current systems — you can't say, like, hey, go do this task for my job, and it goes off and clicks around the internet and calls someone and looks at your files and does it. And without that, it feels definitely short of it.”

1998–2026

“Think for a moment that I granted you the right to buy 10% of one of your classmates for the rest of his or her lifetime. Now, you can't pick one with a rich father, that doesn't count. I mean, you've got to pick them, pick somebody who's going to do it on their own merit. And I gave you an hour to think about it. Which one are you going to pick among all your classmates as the one you want to own 10% of for the rest of their lifetime? And are you going to give them an IQ test? Pick the one with the highest IQ? I doubt it. Are you going to pick the one with the best grades? I doubt it. You're not even going to pick the most energetic one necessarily or the one who displays the most initiative. But you're going to start looking for qualitative factors in addition to, cuz everybody's got enough brain and enough energy. And I would say that if you thought about it for an hour, decided who you're going to place that bet on, you'd probably pick the one who you responded the best to because the one that was going to have the leadership qualities, the ones who are going to be able to get other people to carry out their interests. And that would be the person who was generous and honest and gave credit to other people even if for their own ideas, all kinds of qualities like that.”

2023–2026

“So one is very high in what's called trait openness, which is one of the Big Five, which is basically just like, flat out open to new ideas. And of course, the nature of trait openness is trait openness means you're not just open to new ideas in one category, you're open to many different kinds of new ideas. And so we might talk about the fact that a lot of innovators also are very creative people in other aspects of their lives, even outside of their specific creative domain. So that's important. But of course, just being open is not sufficient, because if you're just open, you could just be curious and explore and spend your entire life reading and doing, talking to people and never actually create something. So you also need a couple of other things. You need a high level of conscientiousness, which is another one of the Big Five. You need somebody who's really willing to apply themselves, and in our world, typically over a period of many years to be able to accomplish something great. They typically work very hard. That often gets obscured because the stories that end up getting told about these people are, it's just like this kid, and he just had this idea, and it was like a stroke of genius. And it was like a moment in time and was just like, oh, he was so lucky. And it's like, no, for most of these people, it's years and years and years of applied effort. And so you need somebody with an extreme, basically, willingness to defer gratification and really apply themselves to a specific thing for a long time. And of course, this is why there aren't very many of these people, there aren't many people who are high in openness and high in conscientiousness because to a certain extent, they're opposed traits. And so you need somebody who has both of those. Third is you need somebody high in disagreeableness, which is the third of the Big Five. So you need somebody who's just basically ornery, because if they're not ornery, then they'll be talked out of their ideas by people who will be like, oh, well, because the reaction, most people have new ideas is, oh, that's dumb. And so somebody who's too agreeable will be easily dissuaded to not pursue, not pulling the thread anymore. So you need somebody highly disagreeable.”

2013–2023 · changed view

“Please mishandle. So something that is fragile, if you map it properly mathematically, you'd realize that the fragile is what does not like disorder. It doesn't want mishandling, it doesn't want volatility, it wants peace and predictability. So the opposite of that would be something that love volatility.”

2019–2026

“And certainly the Girardian intuition is that one of the you know one of the things is always that you know there's this really big problem of violence and scapegoating that in some ways some sense Judaism and then Christianity it's the same story It's the same story of you know sacrifice but it's told not from the point of view of the violent community It's told from the point of view of the innocent victim And there's a certain way where it sets in process this gradual this dynamic revelation that has that leads to the sort of gradual unraveling And there are and as you stop believing in scapegoats you're forced to come up with other explanations and that includes science.”

2022–2026 · changed view

“The consistent thing I've been always interested in, for me at least, is AI, and that's probably where I'm spending my rest of my life on, because I just care about a lot. And I actually care about like many other problems as well, like say aging, which I basically view as disease, and I care about that as well, but I don't think it's a good idea to go after it specifically. I don't actually think that humans will be able to come up with the answer. I think the correct thing to do is to ignore those problems and you solve AI and then use that to solve everything else. And I think there's a chance that this will work; I think it's a very high chance. And that's kind of like the way I'm betting.”

2024–2026

“Looking at the stripe data, one thing I will say that I think is germane to people here. There are many more businesses getting started now than there were a year ago. Like as little as a year ago. Way, way more than we're getting started five years ago. And actually the relative change between last year and this year is pretty much the largest relative change we've seen in any given year. So for example, from 19, from 2019 to 2020, we saw a big jump, you know, understandable during COVID so February to April of 2020 or whatever, you know, I think the growth rate inflected to maybe 50% or thereabouts year over year. In terms of new businesses getting started. As I speak, the number of new business starting on stripe is up around, it's a bit under, but around 2x year over year, which again is the largest relative jump we've seen.”

2021–2026 · changed view

“The times ahead will be radically different from those that we've experienced in our lifetimes, though similar to many times before. How do I know that? Because they always have been. Over my roughly 50 years of global macroeconomic investing, I've learned the hard way that the most important events that surprised me, did so because they never happened in my lifetime. These painful surprises led me to study the last 500 years of history for similar situations where I saw that they had indeed happened many times before with the ups and the downs of the Dutch, British, and US empires. And every time they did, it was a sign of the changing world order.”

2023–2025 · changed view

“For me it's always been about autonomy. I just be able, I want to be able to do what I find interesting, and that's been the core of it. And what I do will change, but the core of having the freedom to do it has been the center of it. And so for me I don't think my freedom has, I mean, not in recent history, my freedom hasn't fundamentally changed in any way. And so I would say that I'm the same level of contentedness as I was last year, but I find engagement in what I do, and that's, I'll give you my definition of happiness, which is doing what you like to do with people you like and doing that as much as you possibly can, and that's my simple definition.”

2020–2026 · changed view

“When we decide to focus what the brain really does is it switches on a set of circuits then all the frontal cortex and nucleus basalis and some others and it's trying to understand duration, how long something's gonna last, path, what's gonna happen, and outcome, what ultimately is gonna happen. So duration, path and outcome. You know, the events of early 2020 are a good example of this. One of the reasons why it's so exhausting to be alive in 2020, is because we are now having to pay attention to duration, path and outcome. How long is this thing gonna last? When are, you know, when are they gonna open up all businesses? Did I touch that door handle? Does it matter? You know, who are the experts? Are there any experts? You know, there are a lot of questions whereas normally, we can just move through life without having to do all that analysis.”

2024–2025

“Now the thing that tells us conclusively that there are parallel universes, wait for it, is that if you look at that, there are certain points on the screen where photons — let's say you have one slit open — you have where photons are arriving, and if you open the second slit, they stop arriving there. So how come? Well, obviously they must stop arriving because they know that there are two slits open now and not just one. And if you look at the mathematics of this, what determines whether they arrive or not depends on whether — how many slits are open there, 1, 2, 5, 17, they all give a different pattern, and the mathematics tells us when opening a new slit will stop light arriving at a given point.”

2015–2025 · changed view

“And so busy mind can often rob you of peace of mind, because what— the peace that we seek is not peace of mind, it's peace from mind, right? So if you look at all the crazy activities you do to be happy, right, whether it's like trying to get laid and have an orgasm, or, you know, extreme sports, or looking at something beautiful, or taking a psychedelic, you're trying to get out of your own mind. You're trying to get your monkey mind to stop chattering at you for a moment. You're trying to get peace from the mind. And there are other better ways to do that. Most of the ways we try to get peace from mind are indirect, whereas if you understand things, if you see things properly, you will naturally slowly develop peace from mind.”

2025–2026

“And so, we never actually had on our radar psychedelics. They were always either an ancient medicine being used in ritualistic practices or being pointed at things like depression and anxiety in certain trials. But it was never understood as a rejuvenation protocol, something that was for anti-aging.”

2021–2025

“One has to understand that when they come into my domain, what they're coming into — they're coming into a very biased place. And my bias is holistically and directly towards ultimateness, truth, world class, reality, the whole thing, the peak of Everest, the ultimate of the ultimate, the ultimate possibility of a human organism. That's where — that's everything I talk about is from there. So I don't have any space or any interest, or I make no room for commoners who seek common things, who seek a little bit of this and a little bit of that. And that's perfectly fine, but that's not what I'm interested in. I don't care about that. And so therefore, could you meditate and get something out of it? Yes, you can. If you meditate, you can get some peace for 20 minutes a day, sure. You commit any crimes? Of course not. You know, that's not a crime. So — but I'm looking at everything from Buddhahood. I'm looking at everything from the ultimate possibility of a human being, both in profession, in skill development, in learning, in life, in internal freedom, everything. The ultimate is the only thing which interests me. I'm not interested in the low and the middle of the road. I'm not interested in dabbling.”

2021–2025

“Here's a mechanistic explanation for why that happens, okay? So visualize NYT and centralized media as a hub, okay? And then here's a spoke coming off, and this is, let's say, Bitcoin, and then it's another one that's like India, and here's another one that's like Japan or Iraq or whatever, right? These various spokes. And if you are an expert on Bitcoin, you have better local information on that topic, in, like, the Hayekian sense, but not simply about transactions, just other things, events that are happening. And so you have a better source of information than the centralized hub, and so you know that what they're saying is, is false about this topic or it's misleading. But then when you go and read about, uh, you know, India or Japan or Iraq or Hollywood or what have you, you do not have, you're not in that spoke, right? So you don't have local information there, you're relying on the hub, which is like a, you know, a hall of mirrors that's giving you a noisy reflection of what's happening in Japan or what's happening in India, okay? It's reflecting back to you something that's false, but you don't have a direct peer-to-peer connection with India, with Japan, with Hollywood, to get better information.”

2019–2026 · changed view

“My basic rule is to read as much as possible. A lot of books are sent to my house, so I order on Amazon. On a weekday I might get five to ten review copies, and I look at each and every one, and I read some of those. I love to reread classics. The earlier parts of my life, I spent a much higher percentage of my reading reading the kinds of books that would be in the back section of Harold Bloom's The Western Canon. Most of those I've read, say, two to five times. And then I read what my friends write. I spend a great amount of time on Twitter, and I love to print out, say, economics working papers and read those. And then I take books on trips. I'm about to take a trip, and I just keep on reading, basically.”

2007–2023 · changed view

“What are the core ideas that have helped me? Well, luckily I got at a very early age the idea that the safest way to try and get what you want is to try and deserve what you want. It's such a simple idea. It's the golden rule, so to speak. You want to deliver to the world what you would buy if you were on the other end. There is no ethos in my opinion that is better for any lawyer or any other person to have. By and large the people who had this ethos win in life, and they don't win just money, just honors and emoluments. They win the respect, the deserved trust of the people they deal with. And there is huge pleasure in life to be obtained from getting deserved trust. And so, and the way to get it is to deliver what you'd want to buy if the circumstances were reversed.”

2024–2026

“Well, you can really think of the economy as digital and physical intelligence. So what we have right now advancing very rapidly is digital intelligence. Obviously with AI, that is, it seems almost with every passing week there are more AI breakthroughs. It can do something amazing that it couldn't do before. But it's still somewhat confined to the digital realm. It doesn't have end effectors. You need the end effectors in the form of humanoid robotics. Human, you need a basically a lots of bots, in a nutshell. You need lots of bots. And then you can go from intelligence manifesting itself only digitally to shaping atoms. And you say, what is an economy? An economy is the production of goods and provision of services. And if you have vast numbers of robots with vast amounts of digital intelligence, you have a sort of a quasi infinite economy. So this is really, the future is going to be very, very different from the past. There's no analogy or metaphor that I think illustrates the magnitude of change that we're going to experience here.”

2023–2026

“That moment in time for our company gave us so much confidence. And the reason for that is you can succeed in doing something, inventing a future, even if you were not informed about it at all. And it's kind of my attitude about everything now. When somebody tells me about something and I've never heard of it before, or if I've heard of it, don't understand how it works at all, my first thought is always, how hard can it be? And it's probably just a textbook away. You're probably one arXiv paper away from figuring this out. And so I spent a lot of time reading arXiv papers, and it's true.”

2023–2026

“For a lot of people, not everybody, but for a lot of people, the more you are showing off or you just desire to show off, a lot of it is a wound that was inflicted upon you and you're like, I'm going to get back. And so, and sometimes too, you're not even showing off for other people. You're doing it for, you're doing it to prove to yourself.”

2023–2026

“When I was a kid, I assumed that there were always some adults in the room. Someone had a plan. Someone knew everything that was going to happen. Someone had it all figured out. And I sort of think why people like conspiracy theories is it's nice to think that someone's got a plan. It's nice to think someone that, you know, has it all figured out. And then I got a little bit older and I sort of started to suspect there are no adults in the room. No one, people have plans. I have plans, but no one has all the answers. No one knows where it's all going to go. And now that I am the adult in the room, I can say with certainty, no one knows where it's all going to go. Like, I'm the guy in the room and I have some guesses and I have some plans, and we're working really hard. But like, you know, we try to always say what we think the possibilities are, what we think is most likely. Often we're right. Sometimes it's in the broader set. And sometimes it goes in a totally different direction than anything we thought. And you know, we keep trying to make progress, figure out more. We try to tell people, not just tell, we try to show people by like deploying these systems and say, "You can go use it. Don't just take our word for it. Try it out. See what it can do."”

1998–2020

“Basically, I don't want to buy any stock where if they close the New York Stock Exchange tomorrow for 5 years, I won't be happy owning it. I buy a farm and I don't get a quote on it for 5 years and I'm happy if the farm does okay. You know, I buy an apartment house, don't get a quote on it for 5 years, I'm happy if the apartment house produces the returns that I expect. But people buy a stock and they look at the price the next morning and they decide whether they're doing well or not doing well. That's crazy cuz they're buying a piece of a business. That's what Graham, the most fundamental part of what he taught me, you know. You're not buying a stock, you're buying a part ownership in a business. You will do well if the business does well and if you didn't pay a totally silly price. And that's what it's all about.”

2014–2025 · changed view

“Software is a technology that, it's like a, you know, the concept of the philosopher stone. The philosopher stone in alchemy transmutes lead into gold, and Newton spent 20 years trying to find the philosopher stone, never got there, nobody's ever figured it out. Software is our modern philosopher stone, and in economic terms it transmutes labor into capital, which is like a super interesting thing.”

2008–2024 · changed view

“So there was an expression until we saw Australia, and effectively one, the sighting of a single bird destroyed millennia of confirmation. So it was a post as a logical problem, by showing that there's no reason, you cannot rule out a Black Swan because you haven't seen any. Okay, my problem is not a logical question. My Black Swan is an event, it's not a bird. So, and it's an event that has three properties.”

2024–2026 · changed view

“The history of it was it was maybe anchored on two visions of what AI meant, and one was Nick Bostrom, Oxford prof, who wrote this book Super Intelligence, 2014, and it was basically AI was going to be this super duper intelligent thing, way, way Godlike intelligence, way smarter than any human being. And then there was sort of the, I don't know, the CCP Chinese Communist rebuttal, the Kai-Fu Lee book from 2018, AI Superpowers, I think subtitle something like the race for AI between Silicon Valley and China or something like this. And it, it was sort of, it defined AI as it was fairly low tech. It was just surveillance, you know, facial recognition technology, we would just have this sort of totalitarian Stalinist monitoring. It didn't require very much innovation, it just required that you apply things, and basically the subtext was China is going to win because we have no ethical qualms in China about applying this sort of basic machine learning to sort of measuring or controlling the population. And those were sort of like, say, two extreme competing visions of what AI would mean in the in the 2010s.”

2022–2026

“Yeah, the data engine is what I call the almost biological feeling like process by which you perfect the training sets for these neural networks. So because most of the programming now is in the level of these data sets, and make sure they're large, diverse and clean. Oh, basically you have a data set that you think is good, you train your neural net, you deploy it, and then you observe how well it's performing, and you're trying to always increase the quality of your data set. So you're trying to catch scenarios basically there are basically rare, and it is in these scenarios that the neural nets will typically struggle in, because they weren't told what to do in those rare cases in the data set. But now you can close the loop, because if you can now collect all those at scale, you can then feed them back into the reconstruction process I described and reconstruct the truth in those cases and add it to the data set. And so the whole thing ends up being like a staircase of improvement of perfecting your training set. And you have to go through deployments so that you can mine the parts that are not yet represented well in the data set. So your data set is basically imperfect, it needs to be diverse, it has pockets there are missing, and you need to pad out the pockets. You can sort of think of it that way.”

2022–2024

“In many domains, I don't know. For example I have no facility with or experience with doing things in hardware, which is not a small domain. If you wanted to become a super skilled practitioner there, what's the best career path? I don't know. Maybe it's to drop out and join SpaceX or something. I'm not necessarily endorsing pursuing the most establishment and credential oriented path. People should try to find the gradient of maximal learning in whatever it is they care most about. The question then is what that is. For biology, not that I'm a biologist, but it is very clear that in order to do really good work, there are a lot of "bench" skills one has to acquire and there is a lot of actual specific knowledge. Any kind of life wasn't designed with neat fundamental principles the way that maybe physics was. A lot of it is obviously evolved and contingent and messy and complicated and all the rest. So there is a lot of specific factual stuff to learn. For those two reasons, there are very few successful pure autodidacts in biology. In virtually every case that I'm aware of, at some point, you have to get direct experience in and with a top lab, where you're seeing how people actually do it in practice.”

2018–2019

“What do I mean by an idea meritocracy? I want to be very tangible for you. In order to have an idea meritocracy, in other words, the best ideas winning out, wherever those ideas come from, you have to do three things. The first is that you have to put your honest thoughts on the table, honest thoughts on the table, for everybody to see and then look at. One of the problems of most groups of people, most organizations, most individuals, is that they don't want to be honest, and they don't like disagreement. You can go out to a restaurant and somebody says, "I like the food," somebody is going to be reluctant to say, "I don't like the food." Thoughtful disagreement is the second thing. You have to understand the art of disagreement, because if there's disagreement, that means somebody is going to be wrong, and how do you know the wrong person isn't you? You have to understand the art of thoughtful disagreement. Then once you have that disagreement to raise your probabilities of making better decisions, and that certainly happens if you do it well, then you have to have ways that are going to, idea meritocratic ways, of getting past your disagreement. In other words, you have to have protocols, and we do something which we call believability weighting decision-making based on different people's credibilities. That's what I mean by an idea meritocracy.”

2023–2026

“Whenever I approach something that I don't know how to do, it always feels amorphous. It's like this big thing and I don't know where to start. Like how do I grab onto a piece of it? And so having gone and attacked like these amorphous things like start a business which actually means nothing. You have to break things down to much smaller steps of action. That's why for me the most basic thing to start a business is like get an LLC, open a bank account, have the ability to process money, and then ask a stranger if you can do something for them in exchange for money. As soon as you do those four things, congratulations, you're a business owner.”

2021–2026 · changed view

“Adenosine is a molecule in our nervous system and body that builds up the longer we are awake, so if you've just slept for 8, or 9, or 10 really deep, restful hours, adenosine is gonna be very low in your brain and body. If, however, you've been awake for 10, 15, or more hours, adenosine levels are going to be much higher.”

2022–2026

“So when I say any, I'm — what I mean is that we're not limited in the programs we can run apart from by speed and memory capacity. So all limitations on us, hardware limitations on us, boil down to speed and memory capacity. And both those can be augmented to the level of any other entity that is in the universe. Because, you know, if somebody builds a computer that can think faster than the brain, then we can use that very computer or that very technology to make our thinking go just as fast as that.”

2015–2025 · changed view

“I think reading is so powerful. If you take away one thing from this podcast, just figure out how to read. And I say that because there are many skills and gifts that people have in life, and the great thing about reading is you can use that to pick up any new skill. So if you learn how to learn, it's the ultimate meta skill, and I believe that you can learn how to be healthy, you can learn how to be fit, you can learn how to be happy, you can learn how to have good relationships, you can learn how to be successful. These are all things that can be learned. So if you can learn, that is a trump card, it's an ace, it's a joker, it's a wild card. You can trade it for any other skill.”

2021–2024

“Conditioning by living in the world. And by living in, when I say don't live in the world, I'm not saying you should live in a cave, but what I am saying is that if you were to be locked in a room and you had loud speakers on and speakers right next to your ears, you know, like they do in circumstances of, you know, prisoners of war and torture. You know, that you would hear the same things again and again and that would become your way of being, because that's all you know. And that conditioning is a systematic and intense and incessant deviation and removal of oneself from one's natural self. So from the moment that a human being is born, you know, he's born natural and perfect. And from every single day from that moment, every single influence that he has is one more step in a direction that is out to sea. And after some point he has no idea where they are, who they are because they've just become a stooge for the world.”

2019

“I think to found a company, I cannot speak for everyone, but I think for a large percentage of founders including myself, it is to get something that you cannot buy that you must build. So for example for Elon Musk, why SpaceX? Because he could not buy a trip to Mars, he had to build a trip to Mars, right”

2021–2026 · changed view

“That said, as a philosopher I disagree with her on most things, and I did even, like, as a boy when I was reading her. I read Plato before Rand, and in a Socratic dialogue there's all these different points of view being thrown around, yeah, and whomever it is you agree with, you understand the wisdom is in the coming together of the different points of view. Yeah, and she doesn't have that.”

2020–2026

“Yeah. And now that won't solve for the daily fluctuations, which happen and sometimes are pretty extreme. But over time it would, that is the right way to invest in stocks, because a company is just a group of people assembled to create products and services. So you have to say, what are, how good are those products and services? Are they likely to continue to improve in the future? If so, then you should buy the stock of that company and then don't worry too much about the daily fluctuations.”

2023–2026

“There's long periods of suffering and loneliness and uncertainty and fear and embarrassment and humiliation. all of the feelings that we most not love that creating something from the ground up and Elon will tell you something similar very difficult to invent something new.”

2022–2026 · changed view

“I hate that I'm quoting Buffett again, I don't want to do this ad nauseam for this whole podcast, but he has a great quote on wealth where he says he wants to leave his kids enough money that they can do anything but not so much money that they could do nothing. I think that's really the key, and that's how I think about my own kids, who are very young. But when my wife and I think, like, how do we want to use whatever savings that we have to benefit them, giving them a safety net but not a fuel is a — that's what my parents did for myself and my siblings. We — I always, when I was a teenager and in my early 20s, I always knew they would be there if I fell on my face, and they would — like, I would never just completely fail, I'd never be homeless, I would never — they'd always catch me. But they were never going to be a fuel. They're never just going to give me money just to make my life better. That was never going to be the case. And so knowing that they were going to be there, but knowing that I had to make it on my own, was so important for me. That's what I want to do for my own kids.”

2014–2026 · changed view

“Before we jump in on the how, I want to talk about why you should start a startup. I'm somewhat hesitant to be doing this class at all, because you should never start a startup just for the sake of doing so. There are much easier ways to get rich, and everyone who starts a startup always says — always — that they couldn't have imagined how hard and painful it was going to be. You should only start a startup if you feel compelled by a particular problem, and that you think starting a company is the best way to solve it. The specific passion should come first and the startup second. In fact, all of the big successes we have at YC followed this.”

1998–2026 · changed view

“A good business is the one that earns a high rate of return on tangible assets. Very simple.”

2024–2026

“there's always a question I like to ask, what is the antonym of good? And the moral antonym is evil. The functional antonym is bad. A Ferrari is an evil car because it doesn't get many miles per gallon. A Trabi is a bad car. It works.”

2025

“If it is the case that those tokens are not any more useful to your next query, I encourage you to do this, because these tokens in this window are expensive, and they're expensive in kind of like two ways. Number one, if you have lots of tokens here, then the model can actually find it a little bit distracting. So if this was a lot of tokens, the model might, this is kind of like the working memory of the model, the model might be distracted by all the tokens in the past when it is trying to sample tokens much later on. So it could be distracting and it could actually decrease the accuracy of the model and of its performance. And number two, the more tokens are in the window, the more expensive it is, by a little bit, not by too much but by a little bit, to sample the next token in the sequence. So your model is actually slightly slowing down, it's becoming more expensive to calculate the next token, and the more tokens there are here. And so think of the tokens in the context window as a precious resource. Think of that as the working memory of the model, and don't overload it with irrelevant information and keep it as short as you can, and you can expect that to work faster and slightly better.”

2022–2024

“A kind of rigor and clarity of thought in that I think so many organizations prize sort of smoothness and smoothness of sort of interactions and trying to reduce minimize the number of sort of ruffled feathers and they kind of at least sort of inadvertently if not deliberately prefer cohesion over correctness and we really try to identify people who are seeking correctness and who don't mind being wrong and who are willing to at least contemplate things that seem improbable or surprising if true or really divergent to what is sort of the generally accepted status quo and that's hard to find and I don't think most of the sort of educational institutions that we all tend to have attended actually do a great job of teaching that and so we look for that kind of combination of sort of openness and rigor.”

2013–2025

“Though the economy might seem complex, it works in a simple, mechanical way. It's made up of a few simple parts and a lot of simple transactions that are repeated over and over again a zillion times. These transactions are above all else driven by human nature, and they create 3 main forces that drive the economy. Number 1: Productivity growth. Number 2: The Short term debt cycle. And Number 3: The Long term debt cycle.”

2023–2026 · changed view

“People will take a finite game, where you have known only known players, agreed upon rules, and a set outcome for winning, and they will apply it to an infinite game. So people are like, I want to, they apply a finite contract to health, they're like, I want to win health. It's like, you don't win health. Like, okay, you're in shape now what? You stay in shape, you keep staying in shape. I want to win at marriage. You don't win at marriage, you keep the marriage going, that's playing the game of marriage if you want to. You don't win at business, you keep playing the game of business. And so we want to take these finite contracts and put them on infinite games, and I think that's where people get in trouble because they're like, I have to keep moving the goal post. But if the goal post is to play, then you win by playing.”

2022–2026

“I eat the same way I have for the last 25, 30 years. I try and get some protein in every meal, high quality protein. I love fruits and vegetables. I like high quality fats, so olive oil, a little bit of butter, a little bit of Parmesan cheese here and there. I don't get into the whole seed oil mess. I think you can lose neurons getting in that debate, just olive oil. Nobody debates it. A little bit of butter. That's fine with me. I do eat starches, so I'm an omnivore. I will have some bread, rye bread, sourdough bread, oatmeal, rice. Depends on how much I'm training. If I'm training more, I eat more of that stuff. If I train less, I tend to eat a little bit less of that stuff.”

2021–2026

“That's right. That is what allows human type knowledge to have this universal reach, whereas biological knowledge, although it has enormous reach, as you said, it can make elephants trunks and giraffes necks and so on, it is actually limited because of the type of knowledge that is. For example, to evolve from one animal to another over, let's say, a thousand generations, every single generation must be viable as an animal, because the DNA strands have to be able to just replicate in every single generation, and if one of them fails, if one of them doesn't work, then the species will go extinct. Whereas humans, as the philosopher Karl Popper said, we can let our ideas die in our place, and that means that we can go through a whole bunch of — a sequence of ideas that aren't viable on the way from one viable idea to the next. We don't have to die when they're wrong.”

2019–2025 · changed view

“If you want to make money, you have to get paid at scale. And why you? That's accountability. At scale? That's leverage. And just you getting paid, as opposed to somebody else getting paid — that's specific knowledge.”

2021

“A person needs his spouse to love him, he needs his children to revere and respect him, he needs to be loved by his children, he needs to be told that he's doing a good job, he needs the adoration of his friends, he needs the companionship of various different people, he needs the validation of the world. The needs spread so far and so wide, you know, they span the earth 16 times. How in the world would you begin to prescribe your way out of that? I mean, that's just it's beyond unserious.”

2024

“And it shows that despite all the technological progress we've had in computers some Force has been not just holding it back but reversing it in in drug development which is part of the reason that drugs have gotten so expensive Eroom's law is Moore's law spelled backwards.”

2026

“It does not understand. It's not conscious. It's not sentient. I'm pretty sure of that. I know some people have suggested it might now currently be conscious. I find that very very difficult to give credence to. It's a very powerful cash register in a way and the cash register is not conscious.”

2025–2026

“The United Kingdom was where the industrial revolution was, if you will, invented when they realized that they can turn steam and such into energy into electricity. All of that was invented largely in Europe and the United States capitalized on it. We were the ones that learned from it. We industrialized it. We diffused it faster than anybody in Europe. They were all stuck in discussions about policy and jobs and disruptions. Meanwhile, the United States was forming. We just took the technology and ran with it. And so I think we were always in a bit of a technology race. World War II was a technology race. Manhattan Project was a technology race. We've been in the technology race ever since during the Cold War. I think we're still in a technology race. It is probably the single most important race. It is the technology is it gives you superpowers. you know whether it's information superpowers or energy superpowers or military superpowers is all founded in technology and so technology leadership is really important.”

2022–2026

“There are a lot of people that will define risk differently, by and large based off of the time horizon. So if you were to ask me, if the stock market fell 20% and stayed there for three years, is that a risk? A lot of people would say, yeah, that's a huge risk, that's an enormous, like, catastrophic risk. I personally would say that's not a risk at all, and the reason is because my goals lie 50 years in the future, so what's gonna happen in the next three years, I could honestly care less. That's really how I think. But that would not be my answer if I was a 90-year-old retired widow on a fixed income; then I'd have a very different answer for you. So I think, summing all that up, risk is just the odds that something will prevent you from achieving your goals. But the nuance is that everyone has very different goals and aspirations and time horizons, so everyone thinks about risk differently. The takeaway from that is most investing debates where people are arguing with each other — is this a risk, is that a risk, should I buy this stock, is the market going to go up next week — by and large those debates are not actually debates. It's people with different risk tolerances and different time horizons talking over each other.”

2023–2026 · changed view

“We started off just thinking we were gonna be a research lab and having no idea about how this technology was gonna go. Because it was only seven or eight years ago, it's hard to go back and really remember what it was like then. But before language models were a big deal, this was before we had any idea about an API or selling access to a chat bot. It was before we had any idea we were gonna productize at all. So we're like we're just gonna try to do research and we don't really know what we're gonna do with that. I think with many new fundamentally new things, you start fumbling through the dark and you make some assumptions, most of which turn out to be wrong. And then it became clear that we were going to need to do different things and also have huge amounts more capital. So we said, "Okay, well, the structure doesn't quite work for that. How do we patch the structure?" And then you patch it again and patch it again and you end up with something that does look kind of eyebrow raising to say the least. But we got here gradually with I think reasonable decisions at each point along the way and doesn't mean I wouldn't do it totally differently if we could go back now with an oracle, but you don't get the oracle at the time.”

2024

“it started as a two-front war against both excessive dogmatism and excessive skepticism. And so if you are over in the 17th and 18th century a scientist was a heterodox thinker who didn't believe in, let's say, the decayed Aristotelian scholasticism of the Catholic church and you were maybe were empirical and there were these dogmas that you were open to questioning but you also couldn't be extremely skeptical. So if you, you know, can't trust your senses and if I don't believe you're sitting there, extreme skepticism is also incompatible with science. So extreme dogmatism was incompatible, extreme skepticism is incompatible, and the problem, problem is, yeah, it's easy to be against one but if you're, you know, if you're always just against dogmatism then maybe you're too skeptical of everything and if you're always against skepticism, you know, maybe you're too dogmatic. And so there's this very complicated balance where we need to be both anti-dogmatic and anti- skeptical”

2022–2026

“I wouldn't say that setting impossible goals exactly is a good idea, but I think setting very ambitious goals is a good idea. I think there's a— what I call sublinear scaling of difficulty, which means that 10x problems are not 10x hard. Usually 10x harder problem is like 2 or 3x harder to execute on. Because if you want to actually, like, if you want to improve the system by 10%, it costs some amount of work, and if you want to 10x improve the system, it doesn't cost, you know, 100x amount of the work. And it's because you fundamentally change the approach, and it— if you start with that constraint, then some approaches are obviously dumb and not going to work, and it forces you to reevaluate. And I think it's a very interesting way of approaching problem solving.”

2021–2024 · changed view

“Make people an offer so good they feel stupid saying no.”

2023–2024

“So I view the brain as obviously a bunch of cells and parts, etc., but I distill it down to some basic features. First of all, we have an autonomic physiology, I think we'd both agree on that, that regulates our sleepiness and wakefulness, our breathing, our heart rate, stuff that runs in the background. And then we have a lot of circuitry devoted to what I would call impulses, things that we desire, we want to move toward, repetitive behaviors, and we also have some impulses to avoid things that are putrid, painful, etc.”

2017–2025 · changed view

“We have this idea that everything can be taught — everything can be taught in school — and it's not true that everything can be taught. In fact, the most interesting things cannot be taught, but everything can be learned. And very often that learning either comes from some innate characteristics in your DNA, or it could be through your childhood where you learn soft skills which are very, very hard to teach later on in life, or it's something that is brand-new, so nobody else knows how to do it either.”

2019–2026

“List of the top 10 companies by market cap almost completely turns over. Procedures for tenure on the top 50 research universities almost exactly the same. Whatever you think of those, there's something gone wrong in the sector. There's not enough experimentation with how you reward people. More schools should experiment with a different kind of tenure, or reward people more on the basis of practical impact. And again, you might object to any particular solution, but the extent to which experimentation has died at the institutional level, to me is striking.”

2022–2026

“I think what's closer to the truth is that it's rigged against professionals who have to compete against each other on a 90-day clock. What were your returns last quarter and how do they rank next to your peers? That's what you live and die by. Like the rigged game the game's rigged against them. But for ordinary people who are like, I'm just going to invest in my 401k and leave it alone for 30 years. I don't even know what my password is. I don't care. I don't know or care if somebody else outperformed me last quarter. I'm just going to do this and follow my goals. That's where the money's made. And so I think that's the irony is we think it's rigged against you when it's rigged against the pros.”

2025

“No, and I don't think they seem alive. But I understand where that comes from. They don't do anything unless you ask. They're like just sitting there kind of waiting. They don't have a sense of agency or autonomy. The more you use them, I think the more the kind of illusion breaks. But they are incredibly useful.”

2022–2025 · changed view

“And also like the origin of life didn't— was actually fairly fast after formation of Earth. If I remember correctly, just a few hundred million years or something like that after basically when it was possible, life actually arose. And so that makes me feel like that is not the constraint, that is not the limiting variable, and that life should actually be fairly common. And then it, you know, where the drop-offs are is very interesting to think about. I currently think that there's no major drop-offs basically, and so there should be quite a lot of life. And basically, what it— where that brings me to then is the only way to reconcile the fact that we haven't found anyone and so on is that we just can't— we can't see them, we can't observe them.”

2024–2026

“Let's start with the first one, which is nobody knows you exist. And if you're under $1 million in revenue, I can virtually guarantee you that basically everyone on Earth doesn't know you exist. And so, the first 4 hours of every day should be dedicated to solving that problem and going from obscurity to being aware.”

2022–2023

“Here's what's fascinating: those same neural circuits are repurposed for romantic attachment in adult life. The same circuits, which shouldn't surprise us. I mean, why would the brain throw away valuable circuitry? But this whole Freudian notion that, you know, childhood attachment styles map onto adult attachment styles, that's real, that's physiological. Now, one important point, it's not one for one, in the sense that let's say you had a secure attachment to your father, let's say it's a young, a young girl, and as a baby and young child she had a secure attachment to her father and an insecure attachment to her mother. In adulthood, and let's say she's heterosexual, so in adulthood she prefers men as romantic partners. This girl grows up and you might say, well, she had a good relationship to her dad, so she's going to have a good secure attachment style in her adult heterosexual relationships. Ah, often it's not the case. They will transplant or superimpose the insecure attachment style to the mother onto male relationships, but have great relationships to female friends, for instance. So we have to be a little careful to not map one for one. That's important.”

2019–2025

“So, you're basically saying this, unless the laws of physics explicitly forbid it, we can figure it out. And if we can figure it out, we can build it. It may take more time, it may take more resources, but humans are sort of universal explainers, we're universal computers, and to date we have found no law of physics that is a barrier and says that things are not understandable. So, we should be optimistic.”

2026

“There will be jobs for the foreseeable future. So just the energy sector will need to grow tremendously to accommodate all of the compute we need for AI, health care, biomedical trials, taking care of old people as we age, entertainment. People still want creativity from other humans, not just from the AIs. They might want both, but they'll still want there to be, you know, a Taylor Swift, a Zakir Hussain, a Magnus Carlsen.”

2022–2026

“It's true for almost anything in life that the best story wins. Not the best idea, not the right idea. It's just the person who can tell a good story that gets people nodding along, that's who wins. That's been, that's true in politics. It's true in business. It's true in everything.”

2023

“It is what we call a language model because it models the sequence of words or characters or tokens more generally, and it knows how sort of words follow each other in English language. And so from its perspective what it is doing is it is completing the sequence. So I give it the start of a sequence and it completes the sequence with the outcome. And so it's a language model in that sense.”

2024–2026

“almost all decisions that you make in the beginning you have incomplete data, and you have to make decisions anyways. And so it's growing comfortable with taking your best bad guess and being directionally correct rather than searching for a perfect answer, because a perfect answer assumes perfect information, which you can only have after you begin. And so in some ways making a decision is the perfect answer, so that you can get the information feed to then improve the quality of the decision later. And I think that one loop is what a lot of people miss out on, is they spend, they obsess for years sometimes on the perfect pick, the perfect business, the perfect job, the perfect mate, when most of the times beginning each step illuminates the next step, which means the information, the feedback that you get from walking gives you more about where to walk than trying to sit at the beginning in the darkness and pick a direction.”

2022–2026

“I never wanted to be really big, you know, like huge neck or giant arms or anything. I just want it to be strong, lean, some size, but nothing extreme. I'm not about the extremes and I plan to try and carry that theme forward into my 60s, 70s, 80s. I think I'll make it to 100. I don't have a whole lot of interest in going past 100, but who knows?”

2019–2026 · changed view

“People who are talking about AI automating programming have never really written serious code. Coding is thinking. It's automatic structure thinking. An AI that can program as well or better than humans is an AI that just took over the world. That's end game. That's the end of the human species. And I can give you arguments why I don't think that's coming either. People who are thinking— and I know I take the opposite side from some very famous people in this debate, but we're nowhere near close to general AI. Not in our lifetimes. You don't have to worry about it, even in our lifetimes, really. It's so overblown. It's another— it's a combination of Cassandra complex, you know, it's fun to talk about the end of the world, combined with a god complex, like people who have lost religion, so they're looking for meaning in some kind of end of history.”

2025

“In the task of predicting the next token in the sequence, the model has learned a lot about the world, and it has stored all that knowledge in the parameters of the network. So remember that our text looked like this, right, internet web pages, and now all of this is sort of compressed in the weights of the network. So you can think of these 405 billion parameters is a kind of compression of the internet. You can think of the 405 billion parameters is kind of like a zip file, but it's not a lossless compression, it's a lossy compression.”

2024–2025

“So I Define learning by same condition new behavior and so when you go to a video game and you battle through the level and you battle the boss if you keep doing the same thing to the boss and you keep losing then you have not learned because you have the same condition and the same behavior and so I often say that like for anyone who's listening to this podcast if the goal is to get better and you're like man I really want to learn something from this podcast if you listen to this podcast and then you're in the same exact conditions as you were before and then you do not change your behavior you learned nothing and so using that definition has at least allowed me to change my behavior faster which then goes into rate of learning which I Define as intelligence and so a lot of people are like man he's so smart but he just doesn't it's like well then if he doesn't change his behavior and he's in the same conditions he's a dummy he's not that smart and so if you are trying to battle the same boss over and over again and you don't change what you're doing and the boss keeps beating you then it's not the game's problem it's your problem you are the problem.”

2022

“Now, the important thing to understand is that it is the poison, the acetaldehyde itself, that leads to the effect of being inebriated or drunk. I think most people don't realize that, that being drunk is actually a poison-induced disruption in the way that your neural circuits work.”

2019–2025

“There's a view, a pervasive view, maybe even dominant view globally, that you raise the children for society, not for the parents. So it's fundamentally a freedom, pro-American thing to raise the kids for themselves, is the next step. So enlightened society would go from 'we're raising the kids for the state' to 'we're raising the kids for the parents' to finally 'we're raising the kids for themselves.' Or we're just not even raising the kids; we're there to help them raise themselves.”

2025

“We are not in a good position to create these token sequences for the LM, and they're useful by imitation to initialize the system, but we really want the LLM to discover the token sequences that work for it. We need to find, it needs to find for itself what token sequence reliably gets to the answer given the prompt, and it needs to discover that in the process of reinforcement learning and of trial and error.”

2023–2024

“If you define work, at least the way I do, which is output, so I define work by output, and in order to get output it's volume times leverage. So how many times you do the thing times how much you get for each time you do it. And so that is the, do you work smart or do you work hard, it's you do both. You do as many reps as you possibly can and you are, you do it with the most leverage possible. So if I make 100 phone calls, the leverage that I can have there would be how skilled I am. So if I make 100 calls I might get 10 times more, and so I worked more, I had more output than somebody who has less skill, but the only way you get skilled is by doing more inputs, by working more. And so it's this virtuous cycle of doing more and getting better, and then you get more for what you do.”

2022

“What are neural nets? Neural nets are these mathematical expressions, fairly simple mathematical expressions in the case of multi-layer perceptron, that take input as the data and they take input the weights and the parameters of the neural net. Mathematical expression for the forward pass followed by a loss function, and the loss function tries to measure the accuracy of the predictions, and usually the loss will be low when your predictions are matching your targets, or where the network is basically behaving well. So we we manipulate the loss function so that when the loss is low the network is doing what you want it to do on your problem, and then we backward the loss, use backpropagation to get the gradient, and then we know how to tune all the parameters to decrease the loss locally. But then we have to iterate that process many times in what's called the gradient descent. So we simply follow the gradient information and that minimizes the loss, and the loss is arranged so that when the loss is minimized the network is doing what you want it to do. And yeah, so we just have a blob of neural stuff and we can make it do arbitrary things, and that's what gives neural nets their power.”

2023

“But if we define leverage as the difference between what you put in and what you get out, so if you have a lot of leverage, you put a little bit in, you get a lot out. If you have no leverage or low leverage, you put a lot in, you get a little bit out. And a lot of times people who are listening to this and are not making as much money as they want, they're putting lots of input in and not getting a lot out. They have low leverage opportunities. And so understanding how to get more for what you put in is the game overall.”