AI CEO Speaks Out On the Dangers of AI (And How to Win Despite It All): Brendan McCord

The stuff we're building now, it mediates 20% of waking life. What we think we know about AI today doesn't necessarily hold for the future.

I think in this decade, for the first time ever, anyone who can think clearly, take action and communicate their ideas will get rich. But also, we may lose more jobs than ever before, be replaced by robots and AI in ways we can't even fathom now and potentially leave an entire generation behind. I think it's A Tale of Two Worlds. So today I'm bringing you Brendan McCord, one of the sharpest minds in AI. What will happen to our human need to labor?

I think more jobs will arise out of this growth, but there'll be unpredictable categories. It's an unknowable future.

So we've become this nation that used to be owners. Now less than 10% of us own a business. And maybe there is a way that AI can help us take that back.

Will people use artificial intelligence to realize a new kind of independence? Alternatively, will people use it as an autocomplete for life? Will they basically say, hey, tell me what to do? If we become a nation of order takers or sheep from artificial intelligence, it will kind of solidify this trend, whereas this is our moment to reverse it.

But we're almost like a little puppet on the strings in some ways.

Exactly.

That's so fascinating. Hi, I'm Cody Sanchez, and this is the Big Deal podcast. Today I'm bringing you Brendan McCord, one of the sharpest minds in AI. Brendan's not just thinking about where AI is going, he's thinking about where we're going and how to stay human while we get there. He built not one, but two AI startups that got acquired for $400 million. And prior to that, he was deep within the government, where he was a founding chief architect at the Joint AI center for the U.S. department of Defense. He also was part of the DoD's first applied AI organizations, and he led the strategy, so both classified and public versions, to figure out how to do machine learning at the Department of Defense. Now he's working at Cosmos Institute, which he founded. And I believe he's attempting to train the next generation of philosopher builders who believe in humans and that AI should support us, not the other way around. This podcast is going to help tackle the question of what will AI kill first? Can we make money with AI and do humans still thrive? It is a big mission. So this episode today, I think, could change your life. If you listen close closely, it's about tech, sure, but it's also about power, autonomy and building a world worth living in. And so I wanted to have you on to talk about AI because it's been dominated in my mind lately and in a very like neophyte, normal human. It is not part of my day to day that I think about AI. I own Laundromats for sake, you know, so like this is not core. And yet even in my business I feel this visceral fright and also excitement about what is happening in the age of AI. Microsoft fired 7,000 people. I think. Recently the CEO of Upwork sent a note that I thought was fascinating to all of his employees. I don't know if you read it. And it basically says it does not matter if you are a programmer, designer, product manager, data scientist, lawyer, customer rep, or a finance person, AI is coming for you. If you don't make that move, you're going to be out of work. Not only in our company, but also across the industry. There's not going to be demand for people who are working like there were five years ago. So my question to you is what is going to happen with AI and jobs and what will happen to our human need to labor with AI?

So I think like, if you're a CEO, you've got to ready your team for like, for the future. Like as you see it, you don't know, I don't know, like where the thing, where, where things are headed. It's an unknowable future, but you've got to make an attempt and then get your team focused on. And so you got to jolt them sometimes. And I see it in some of those letters where it's like, hey, this job won't exist. But I think underneath that is probably a lot more positive story. In other words, like, you sort of have to separate like the work of the CEO to like rest attention to this and say like, you know, stop doing the thing the way you were doing the thing and start really taking this seriously from the underlying mechanic. And the way I kind of visualize it is like the economy is going to grow and it's going to be this like, you know, growing sphere that on the surface will have all sorts of cracks, you know, and all sorts of like tectonic relocations and shifts and stuff like that. And so I think more jobs will arise out of this growth out of. But there'll be unpredictable categories of jobs. And by now I think people have analogized this a lot to the past. But like a hundred years ago we had an 80% agrarian economy. We were like farming. And if you had told people that we would have a 1% agrarian, you know, 1% were in agriculture, they would have been all kinds of skeptical. They would have said, well, are people not eating grains? It's like, no, we're still eating grains. Are people, you know, just out of a job? Are they just, you know, using leisure? No, people work pretty hard still. I mean, we have a lot more leisure hours. Actually, we do have like four times the leisure hours as we did then. But all these categories of jobs would have arisen that had, you know, like there's a stat that, you know, the beginning of the film industry, you know, would people have been able to imagine that the kind of movie budget of Avatar was like, more than, I don't know, the GDP of the United States? I mean, I may be like a little bit wrong on that, but like the growth is just staggering and weird and we don't realize. But the other, another way to think about it is that like, the bottlenecks, the bottlenecks have shifted over time. They've, they were like muscle in a lot of the earlier economies. They were then like clerical work or repetitious work in others. And now they're things like maybe compute, right? Like when, you know, my friend is building Stargate here actually with Crusoe, and like they're rushing to build this because there's like literally not enough compute.

Explain what Stargate is for anybody that doesn't know.

Yes, Stargate is this like kind of like epic scale project to build massive data center that can train the next generation of AI. The whole idea here is based on just like we had Moore's Law, like people be familiar with Moore's law as a way to describe how chips advance over time, how they get denser and, you know, better. We, we have this thing called scaling laws where AI, you know, we, we can throw more compute at it and we get a bit, you know, we get more performance and we think that we can reliably predict that trajectory. And so what that does is it means that like CFOs and venture investors can invest ahead of that. They can say, look, I don't know what the next generation looks like, but I know we need $1 billion data center and a $10 billion data center and, you know, on and on and on. And so they're building it really quickly and by the time it's built, you know, they're going to need to build a bigger one. Right? And so these, these are new bottlenecks that require a huge amount of jobs, a huge amount of like, labor content. And so figuring out sort of where those bottlenecks are, that's an important thing. The other thing I would say though, about the job is that, you know, it's a good time to go back and think about like a lot of the thinkers at the beginning of the mechanization that happened in the Industrial revolution were, you know, they were very positive on like what automation was, but they thought about, you know, the job as more than just an economic thing. In other words, like, we're not purely economic beings or even primarily like we, we do need work, we do need food, we need like, you know, material comfort. We seek it at least. But like the job, I mean, it's, it's clearly something more about like self architecture or becoming like we, we, we. We develop through the work that we do. And so this is a, I think, mindset shift that probably is necessary to sort of think, think less about the, the narrow, you know, sort of definition of what a job entails.

Yeah. Well, there's two things I want to double tap there. I saw this image the other day that I thought was so fascinating. I actually didn't realize it happened this fast. Cause I was thinking, well, how fast will AI move? Like, how fast do I need to move? What is this angst I need to have on AI? Should I be integrating it? F. And I saw this gent on the Internet share a picture of Fifth Avenue in New York in 1900. And we can, we can put it up here on YouTube and it's like filled with horses in 1900. And then 13 years later, the same Fifth Avenue street has zero horses and is filled with cars. And then I got curious and I was like, well, how many horses were in the US in 1900 versus today? And. And I found it was 21 million horses in the US in 1900 and today somewhere between 6 and 9 million. Obviously we have grown the population, but human population a ton, but we've actually decreased those. Those poor horses replaced with cows that we now eat. And so I guess my question for you is like, maybe it's that same question. It's, it's how fast will this scale and move in relation to the industrial revolution or even just automobiles being created? Like, instead of 13 years, is this three years where all the quote unquote horses are gone and replaced with AI agents?

Yeah, There's a guy who writes about like technological diffusion and he goes back to like the invention of, I'm gonna say the tulip, but I don't Know why a tulip would have to be invented? It's a natural thing. But I think what I mean by that is like the spread of tulips, like tulip growing. I promise. He says that otherwise it wouldn't be in my mind. But he, he writes about all these diffusion curves across. And like one thing you have to keep in mind is the technology, even if it's immediately available, takes a long time to like propagate. In other words, you could stop right now. Google just did its big announcement, like stop, pause, you know, nothing else. And we would have a few decades of like very profitable meaning like good uses of technology to just harvest. Right. And why is this? Well, I think a lot of it is the government. In other words, I think that individuals are able to adapt relatively quickly. Like some things are deeply ingrained. And so my parents will never use A.I. i'm convinced. Right. And it's not because government regulation is getting in the way. But for many of us, we have a kind of plasticity. We've grown up in an environment where we expect change, it's wired, so we can take hold of it pretty quickly. And certainly when we're pushed and incentivized, we do like if we're in a company, if we're fiduciary, it's like, no, we're going to get our lunch eaten. Like, we've got to go. So companies and individuals, especially young ones, want, they're hungry for like the adaptation that it entails. But a lot of things get in the way. A lot of things get in the way and mostly those are interventions based on this, like regulatory scaffolding. I say this because one of the most important things to always keep in mind is in thinking about a society, is the society adaptive? Can it adapt to new technologies, to changed circumstances, to new preferences, open societies, liberal societies in the old sense of the word, they can. Societies that are run by top down mandate, they're highly, highly fragile. And so this is where, you know, U.S. has a distinct advantage versus China versus Europe. Like these are, these are extremely maladaptive. Though they may be more efficient at any given moment to run on a top down basis. It's the, it's the decentralization of the US as much as we've preserved of that, that really renders us like the leaders.

So could there be an argument then that for instance, the US today definitely has more regulations than it did during the 1900s? There's more of a stranglehold overall federally today than there were in the 1900s and so, in fact, maybe usage of AI and proliferation, even though it can go faster, aka you can produce things on the Internet faster than you can an assembly line of cars, it might actually go slower because we have so much regulation.

Yeah, I think there would be. It's. It's like regulation tends to act on the counterfactual. Meaning, like, there's a world in which we. There's a world in which 50 years ago we started using nuclear power. And that world is crazy and awesome. But we didn't. Right. We discovered it. We split the atom, you know, and then we didn't do anything with it. I mean, I did on a submarine, but like, civilian use in the US Was nil, essentially. Right. And so the problem is with these things is, like, you don't get to look at that future. You can imagine it, but it's like we didn't get that, you know, we, we missed the civilizational subsidy that it entailed. So you make a good point. But it's like we have to compare, like, what might have been if we had had fewer restrictions. So, you know, I always try to keep that in mind because the counterfactual is like, Is the thing that matters.

It's a good point. But you do see in Europe right now, for instance, I saw a couple articles this week where they said we are severely behind when it comes to AI because we have sort of too much regulatory burden on people, which basically is just too many laws.

Yeah. And it's cool that Europe can see it. The only reason they can see that is because the US has fewer, and so we're ahead.

That's counterfactual.

But, like, the reality is, yeah, if there's another world, there's this other like us without the regulation, you know, that it does have, that's going even faster and farther. And in that world, we've like, cured cancer, you know, or whatever. We just, we just don't, you know, we don't get to live in that world.

So when you see these headlines that are like, your job will be gone in three years, we will lose, you know, a fourth of the labor force. We will have to instill universal basic income. You know, we will have potentially a recession for the many and proliferation for the few. What do you think? Like, is that wrong? Are they. Are they wrong?

I mean, I think it's wrong on many levels. Yeah. I talked about how I think the, like, there's still going to be bottlenecks. There's still going to be, like, one of the things that's pretty, you know, that humans are extremely good at, particularly entrepreneurs, is like being alert to opportunities that come from interesting recombinations of inputs and like an awareness of preference and of like local circumstance. I know that's a very academic way of describing entrepreneurs. Israel Kersner, scholar, has this like beautiful framing around alertness. But like that never goes away. We need people to be alert to things that humans want and then have the kind of creativity to find and experiment with what they want. Never goes away. As I said before, like the categories of jobs, the people I met here, you know, the former Mr. Beast president, like that job is unintelligible to somebody in the 1980s. Like that doesn't make any sense. That like you would, you would have to sit with somebody for 20 minutes to explain what that job even meant.

Yeah.

You know, and so there's that point on the UBI point. I think that UBI crucially separates like untethers one's contribution from one's reward. And I don't mean this in a kind of moralistic way like we need contributors. I think we do need contributors. But I mean that the price mechanism in markets is the thing that tells us what is worth figuring out about what people value, where the opportunity is. Right. It's like a heat map that goes around and says, look, this profit pool, the only reason the profit pool exists is because there's something valuable there, people value it. And so if you separate, you know, the, the, the, the reward that somebody gets from private property, first of all, you kind of abolish the system of private property or substantially mitigate its usefulness, but you now have people that have no kind of need for directedness towards mutual benefit, towards helping other people. That's a crucial slowdown. That's a guaranteed way to levelize and slow down an economy. You saw this in Britain after the war, by the way. They didn't do UBI exactly. But, but they have a ton of health service type things that basically took a natural distribution of income, meaning that which results from individuals acting freely and trucking, bartering, exchanging. Like Adam Smith would say, you take that distribution and you flatten it and you introduce these kind of National Health Service style things. And what happens is the economy predictably comes to a halt. So I would be much less, I'm much less focused on kind of managing a distribution of income. In other words, saying like, there shouldn't be Elon Musk's and there shouldn't be this thing and like applying a pattern that I think is right Right. I would reject that. I would much more prefer to work on how we create greater wealth that can be distributed, that can be, you know, generated across a society that can help people, you know, who live today and who live. Who will live in the future. Like, the wealth question is the more important one.

Yeah. I also think it's interesting because when you go back to history, there's really not great examples that I can think of. And you are much better on this than I am of when you go back through history and technology came and it eviscerated opportunity and, you know, decreased a population's ability to earn and work. But you have example after example, example of the opposite. And so my case against UBI has always been, well, one, fundamentally, I think that purpose and labor is like, uniquely human and it makes us happy. And that would be like an Arthur Brooks ism that, you know, without that, we're not very happy, in fact. But simultaneously, there don't seem to be very many examples or counter me here. Are there very many examples where technology was introduced and it became worse for society to earn? Now, you could make the argument tech could make us less happy, which you could say social media did in some ways. But, like, has it ever happened through history?

Yeah. So locally it has. Right. So if you have like a group, a community, a region of, I don't know, Indonesia that's set up around the sewing machine, like that's what they do. They produce clothing, textiles, they sew. That's what they've been trained on. And then you have an automated sewing machine that can do that. Then what do we expect? Well, we expect to have dislocation, job loss, that sort of thing. The key there is that this is a. This is a moment for adaptation where it doesn't take these people off of the map. What it means is that the next thing that they need to do is now in view or get that they get kind of motivation to figure out what that next thing is. So we're a highly dynamical system. Like. Like we're constantly adapting. And every opportunity that technology obviates for the individual, meaning everything that technology can handle with automation, simply opens up a new frontier. That's what I hear when I see the CEOs saying, hey, this thing, your job, as you know it, is over the other side of that coin, is like the job you didn't know was needed. The door just opened.

Yeah. And to your point, maybe that that job has 4x more leisure time for you with a higher potential earnings.

Yeah. I mean, empirically it does, right? Like, yeah, we've the lifetime leisure hours have gone from like 40,000 to 160,000 more.

Wow.

And so this is like the marvel of capitalism is that it's given us more leisure.

Most founders think they have a traffic problem, so they throw money at ads, more content, more clicks. But the truth is your site visitors aren't the problem. Your follow up is if someone visits your store, adds to cart, then disappears. Most platforms just let them leave. Omnisend doesn't. It triggers abandoned cart flows, win back emails, and even browse abandonment messages all on autopilot. It's not just email, it's email plus sms. With smart segmentation built in. You can trigger a discount code to someone who clicked but didn't buy. You can follow up with a new product after someone made a purchase. You can even a B test everything to find what actually converts. And you don't need five different apps to do it. OmniSend gives you all the tools without the Frank Frankenstein stack. If you're bleeding leads and calling it low conversion stop. You don't need more eyeballs, you need a machine that closes. That's what OmniSend is. That's so interesting. Like it really needs to be a campaign for that because I think people today think we work more than ever. You know, works incredibly hard. And then you go back and you try to do. I mean we do these series where we go and. And I work in a porta potty business for a day.

Yeah.

I'm like me, this is way harder than, than working on the Internet. Like just categorically it takes longer, it's more work.

Yeah.

And so it is funny that there's not a PR campaign for that and there probably should be.

I mean, I think it's part of the human experience that the things that work well, we are, we are least aware of, we treat them as water. I went to Harvard Business School, which is like a mecca of capitalism or a West Point of. I don't know what to call it, but not West Point, Naval Academy of Capitalism. But there was one class on how markets work and it was called Reimagining Capitalism. And the focus was on what's wrong with it. And it was. The solutions were things that I later came to be aware were things like syndicalist socialism, you know, like worker communes, things like this. It's like, wait a second, we should study the promise of markets. And so I did this. Like, basically I met, I got introduced to Tyler Cowan and who's the chair of MERCATUS It's Latin for markets. He's one of the top political economists of our time. And we created something where we brought entrepreneurs in to read about the fundamentals of markets. Read Theory of Moral Sentiments, wealth of nations from Adam Smith, read these kind of foundational texts, because people don't even understand what markets do. They don't understand how they work. They only kind of criticize things based on, I think, often ill formed conceptions of justice that have arisen around, like the market critics.

Yeah. You know, what's interesting is, I think for young people listening today, or really for anyone, I mean, if you want to break out in your life successfully and you are curious, this is an incredible time with AI to have knowledge at your fingertips.

Golden age.

It's the Golden Age. And also simultaneously, I think it's an incredible moment to matter in history because, I mean, think about where we are in Austin. You've got, you know, University of Austin, this campus that's happening here, where people are trying to do free market principles and think about capitalism in a unique way. Then you've got University of Texas Civitas, Ryan Streeter. Then you've got your Cosmos Institute. There is like this revival, it feels like to me, that is happening of core principles. And I think, you know, for most of my lifetime, those were not highly regarded and, you know, had been under attack in some ways. So I also think it's interesting for young people, like, if you want to find a group of people who really care about laboring and production and who will tell you that, like, when you get lost in that thing that you're so obsessed with that you lose track of time, that that's okay and beautiful. This is like a really cool time to do that.

Yeah.

And, you know, I want to. I guess there's one. Please.

So, actually, a couple of things, but I'll say them quickly. So one is like, I thought of another thing on the reading list. Emmanuel Kant wrote what is Enlightenment? And he channels Horus. I think when he says. He says in Latin, dare to know and have the courage to use your own understanding. This is such a profound thing. Like, this was the Enlightenment project he was reflecting on. We always talk about, like, enlighten the Enlightenment. He writes this, like, definitive reflection on what the Enlightenment was.

And.

And it was like, you know, we previously would be told by our betters what to do in a. In an aristocracy or in a monarchy. You know, now it's like we have the courage to use our own knowledge. You thought, you made me think of this when you thought about, like when we talked about people who are curious using AI systems, like my 3 and 5 year old, it is a golden age for them to figure out. My, my five year old wants to learn gardening. She literally can go to ChatGPT and has and is learning how to become like a gardener. And it's like I'm just blown away. So cool. And then I do want to talk about the Austin scene because I actually do think that's pretty powerful. You know, people will know this, but like that enlightenment thing, it happened in coffee houses in Edinburgh, you know, the, you know, the Greeks met at the Agora. Right. The French, like we talked about salons, now the Parisian salon. And so at every moment there's this like, there seems to be this need to like get together, bump into one another, work on projects. And yeah, the ones you mentioned in Austin are really special. I would say Austin, I'm bullish on its ability to become, you know, not only a great startup hub. I think it's like probably fifth, fifth best startup hub in the, in the, in the US and therefore the world. But, but I think I'm even more bullish on its ability to like fuse the kind of entrepreneurial mindset with deeper thinking about the goals that technology might serve. Like the ends of technology. My pithy way of saying it, because I care a lot about human autonomy, I think this is like a really critical thing that I'm trying to raise awareness on is like, we can be the capital of autonomy. That means autonomous systems, like what Elon Musk is building and all, you know, but also this philosophical conception of what it means to be an autonomous system human being.

Yeah, well, I mean, you're so right. I mean even Silicon Valley, right, that was sort of the center of the Internet for so long and arguably still is in some ways, but in some ways isn't any longer.

Yeah.

And so, yeah, if you want to play a game, go where the game is played. And so I think if you, if you believe in sovereignty and you know, the, the power of the individual and the human, Austin's an incredible place for that. And what's interesting is a lot of those cities, they weren't like major thoroughfares that would become the, the New Yorks and the San Francisco's of the world. I mean, there, you know, Amsterdam, you know, when, when you come to like base level free market principles thinking, you know, that happened in Amsterdam salons too.

Yeah.

And so, you know, it kind of takes me to one of your points that I thought was so interesting. You talked about something that's near and dear to my heart, which is how in the US I believe that we have slowly become this, I say a nation of, of serfs. You say a nation of employees. We've become this nation that used to be owners. 80% of us own things. Now less than 10% of us own a business. And, and maybe that there is a way that I can help us take that back. We can reverse a trend. And those trends are really hard to reverse. I mean since the 1800s we've sort of slowly lost our ownership of businesses. Can you talk about your nation of employees becoming entrepreneurs?

Yeah. So I love this. So Jefferson, his idea for early America was that we would be independent, we would be self reliant, we would be a nation of farmers in a really. And he really cherished, I mean he, he brought into being this idea of, you know, the gentleman farmer. You know, Tolstoy writes about this with Constantine Levin in Anna Karenina. And this is like a beautiful archetype for of like self reliance. But it really was true in a sense for the free, free people within in America that they were self reliant. Like I have my mother, mother and father in law, they have a ranch an hour and a half outside of here. And I grew up in like DC suburbs so I didn't really know what this meant. But like you are very entrepreneur, very self reliant in managing your affairs. There's no one you can appeal to, there's no process you can appeal to. You don't have a boss, right? And so think about a nation like that and what they would have voted for, what their preferences would have been, how they would have governed themselves. Now think about a world in which like 10% of us are self employed entrepreneurs, whatever it is, you know, and most of us are spending all day, every day listening to a boss following a process, kind of turning our mind off that sort of thing. The effect that it has on preferences is really profound. Like Aristotle writes that we are creatures of habit. And this is like a very well known phrase. But if you really take it seriously, like humans are creatures of habit. And so if we are habituated day in and day out to be kind of told what to do, given instructions, here's how you follow the process. It changes us fundamentally. And I'll give you a great example. So people know that east and West Germany grew up under two radically different systems. East Germany, Soviet system, West Germany, more like the West. And in Covid so decades after the fall of the Berlin Wall. You have very different responses among people that grew up under the system of Soviet communism versus the West. They're more pliant, they're more compliant, they're more willing to listen to authority. And I just cite this because it really seems to be the case that the more we are free, the more we are independent, the more we love it. Right. Could you ever go back to being an employee? Hell, no. You could never go back, right? Could I ever go back? I could never do it. I mean, I might, you know, and if I have to serve in government or something like that, but, you know. But you can never do it, right? You just end up loving what this freedom, it plays a huge role in your life. So all this is to say that, like, we find ourselves at a really precarious moment because we've become a nation of employees. Now you bring on artificial intelligence. Will people use artificial intelligence to realize a new kind of independence? I hope so. To, like, raise their potential to be more entrepreneurial? I think that's the hope. Alternatively, will people use it as an autocomplete for life? Will they basically say, hey, tell me what to do? And I. I worry that at the top labs, the AI labs that are really driving this, a lot of people behind closed doors believe that we should in fact do that, that we are being foolish rebels if we don't listen to artificial intelligence. Tell us what to do. They're missing something that's fundamental to human flourishing. You can tell I'm passionate about this. And so if we become, you know, a nation of, like, order takers or sheep from artificial intelligence, it will kind of solidify this trend, whereas this is our moment to reverse it.

Wow, it's so fascinating. You know, it kind of takes me to a point I was thinking about before with you, which is Elon Musk. I heard him say that who develops AI and the principles they follow will decide whether AI becomes the best or the worst thing to happen to humanity. And so what role does human judgment or the people who are determining our AI play in what eventually comes into our hands?

Yeah, I think it's everything. So I. I'll give you some examples, like if you look at past revolutions, so the printing revolution is a big one. And I think about early. We were talking about early America, so I'll stick on that. So Benjamin Franklin was a master printer. You know, he came 200 years after Gutenberg, but he took the printing press, and he took an enlightenment idea that knowledge should not be controlled by the church. It should not be controlled by the state. And he built a lending library and a network of independent publishers.

Yeah.

After Franklin, Joseph Goebbels, the Reich Minister of Propaganda, took the same technology again another 200 years and twisted it into a system of mass manipulation. So propagandized, you know, the country consolidated around the Nazi party vision. This just highlights for me that like the technology was not different. Like what Franklin was doing with his independent publishers and what Goebbels was doing in kind of like corralling all the news. Same technology, right? Enormous difference. One built for freedom, the other built for control. You see this just to give another example, like the inventors of the World Wide Web, of the Internet, Tim Berners Lee, as one example, really had like openness at the core. Decentralization. That was like a major, major philosophical position they went into it with. They wanted the Internet to grow discourse. China, in its great firewall, used that same Internet technology as a way of saying, aha, we can now control what billions of people see. We can create a conformist state that consolidates and sustains power. So I think Elon's right. I think you need to have people that are focused on the human goods, focused on key ideas that underpin the west building these systems. And if you don't have that, you're going to have one of the most perfect systems of control ever, ever delivered to human society.

Hi, this is Deechen Lackman and I am here to tell you Toyotathon is on. So stop by your local Toyota dealer for incredible year end deals on the cars, trucks and SUVs known for their legendary reliability, like the stylish and efficient Camry. The ready for anything rav4. Or check out a rough and rugged Tacoma or Tundra, all with great financing and lease options available to qualified customers. Visit buyatoyota.com to find out more. Toyota. Let's go places.

Rince takes your laundry and hand delivers it to your door, expertly cleaned and folded so you could take the time once spent folding and sorting and waiting to finally pursue a whole new version of you. Like tea time you mm. Or this tea time you. Or even this tea time you said.

You hear about Dave.

Or even tea time tea time tea time you mm. So update on Dave. It's up to you. We'll take the laundry rinse. It's time to be great.

I think that's what why what you're doing at the Cosmos Institute is so important to give you a little plug here. But there's two things I think are fascinating about the Cosmos Institute. One is your speed to action. Like, I have just found with everyone that I've met that is successful, the faster you move, it seems like the more money you make over time. And, like, the fastest people I know are often the most successful. And I think that's counterintuitive because when we're young, we're told, like, be careful, don't take risk, you know, what if you fail? Blah, blah, blah. And yet all of my friends who are quite successful, they move incredibly fast. And so I think. How long has the Cosmos Institute been around now?

We're now like a year and a half.

Okay, so a year and a half from zero idea, basically, to. And then you can explain where you're at today with having done grants for a ton of AI entrepreneurs and kind of coming up with this huge idea of, we want to create the next generation of philosopher builders, let's say. So let's talk about this. And you know, for the people listening, what I want you to think about is, like, when you have like a big, huge, ridiculous idea, how can you move fast on it? And how can you get inspired by somebody who is like, I don't want to just. Not that there's just. But I don't want to just create the next AI company. I want to force, function, change how all AI companies are built.

Yeah.

By providing a platform for everyone and a, A thesis and then rallying some of the biggest names in the world around it.

Yeah.

So can you talk about that a lot?

You should be our pitch person. So what we saw. Yeah, we, we think we need more Franklin's. Right. We need more philosopher builders like Franklin who can take ideas about human flourishing and actually use that to inspire what they build. Right. The stuff we're building now, mind blowing statistic, but it mediates meaning. AI mediates 20% of waking life. So this is very, very serious in terms of being the kind of hidden structure of civilization of human thought.

And by mediates, you mean it takes up about 20% of our waking hours.

Yeah. Or like the information. Yeah, exactly. It guides it. The information you see, you know, it's.

Not always, you don't even realize, but we're almost like a little puppet on the strings, in some ways manipulated by AI for good, bad, or of our own will.

Exactly, exactly. And so it just plays a massive, massive role now and in the future, that number is not going to go down. Right. And so more important than ever, like, this is a moment when philosophy matters and when the builders matter. So we said, okay, that must be so that we need a new kind of technologist. How do we get them right? You look across the landscape and you see universities, almost without exception as being a place where people who are either narrow in their specialization, narrow technician or conforming ideologue, who really is not thinking about these questions very deeply, they're kind of indoctrinated to one to one, you know, view of the world. That's not helpful. So next you look at the tech companies. The tech companies are incredible at building these systems, but they tend to focus more on like the means and not on the ends. Right? Focus on the customer. That stuff's all good, but like on the overall end, how it impacts human life. There's not a lot of, of deliberation there, right. Or if it is, it's all done like kind of privately by the people in tech. And then think tanks are, you know, really smart, but they, they theorize, they don't build. And so we got to build, you know, we realized we got to build a new, a new institution, a new, and we call it an academy. I will just say that there have been moments when, you know, just because in case people think this is crazy, like some institutions have risen to the occasion. Like I mentioned the radiation lab at mit, right? It took, you know, engineers, made them into inventors that helped win World War II or another good example is University of Chicago. It took scholarly economists, turned them into reformers who freed markets across five continents. And so you can do this, like you can create an institution that's fit for the moment. And that's what we're doing at Cosmos. We started a year and a half ago. We got an incredible group of people around it. Tyler Cowen, he's very optimistic about, you know, AI people like Jack Clark, who co founded Anthropic, who's a really subtle thinker about like possibilities of risk and so forth. Some incredible AI researchers that are deeply philosophical, some philosophers who are also technical. So we like kind of have this cool hybrid crew, Very, very special concentration of talent. Just look at them on our website and you'll, you'll see. But we moved out with a kind of like private sector urgency. We're a non profit, you know, we're 501c3. But like, I didn't hire anyone who comes from the non profit world. I, we were super entrepreneurial. I tell people before they come, like, this is not a job. Like this is, this is like we're kind of zealots for the mission. We work extremely hard. And in a year and a half We've backed over 50 projects. We have over 10 fellows. We have an entire new AI lab at Oxford. It's the first in the world to combine philosophy and AI in this way. That lab has shipped open source software. Philosophical paper, an AI paper like it is moving out quickly. And we have put out hundreds now at this point of content pieces, including videos that have gotten over a million views around deep topics in philosophy. Right. We've like, you know, the work we're doing with Jonathan B. If you, you know, he'll sit down for 90 minutes, talk about consciousness with the person who works on that at DeepMind. People are really deeply interested in this. So there's a market demand. And when I get like Twitter DMs and stuff, one of the cool things is that people are recognizing that whereas before in AI people there's like a kind of group that is focused on the doom, the doom scenarios, these effective altruists, existential risk people, rationalists, and then there are the accelerationists at the other end. But Cosmos is really, I think, making progress at COLA at bringing into being a third way that's focused on the human good, you know, on human flourishing, but in a really serious technical sense. So, yeah, it's been, it's been pretty, pretty, pretty wild. Right.

And you guys are giving out million dollar grants too. So if somebody's listening.

Yeah.

And you have a compelling AI idea project. What's the parameters?

Yeah, okay, so make somebody be able.

To get a million dollars from Brendan today.

So. Well, so last week I went to San Francisco. You and I were together in Austin. I flew over to San Francisco for, for a night and we announced. Barry Weiss announced, people know Barry Weiss, that Cosmos is partnering with fire. FIRE is foundation for Individual Rights and Expression. And the, they are the very best when it comes to defending things like free speech. Right. They're what the ACLU could have been, but they're incredibly principled. They're non political, nonpartisan, wonderful organization. We think that the future of free speech, it's about AI. Right. Like I mentioned the 20 stat. So increasingly this is the information, comments and what we have access to, what reaches our mind, what forms in our mind as thought, it's going to be shaped by AI. So FIRE is really, I think, at the forward edge of like leaning into this saying, yeah, this is where the, this is where the thing is headed. So we put together this program to sort of say, look, we need builders to get engaged right now and we need them to build stuff, particularly open source stuff. Right. Because we want to build things that maybe wouldn't have naturally been built at a company. You know, I love it if, like, you know, X AI or Meta or anyone is working on this stuff. I think that's really cool. But we want to kind of motivate the. The independent builders to create systems that promote truth seeking. And what we mean by that is in order to discover truth, in order to kind of have truth matter in our lives, you have to have a clash of ideas. John Stuart Mill tells us this, Chapter two of On Liberty. You've got to have this contestation, this clash. And in order to really think deeply and develop insights that move away from falsity, you have to ask good questions. In both cases. There are problems with AI today in terms of the extent to which you get to see and encounter these minority views or these, you know, provocative views, and in terms of how much it tends to make us want to ask questions. So we want to get ahead of that. And we've already seen, you know, 50 applicants. We just announced it like a couple days ago, but I've seen 50 people apply. There's a lot of demand for this. I think we're going to do a hackathon next week in San Francisco. We're going to be out there a lot. We'll do some in Austin. So we're just. We're building a movement around this because AI is. Is the place where free thought and expression, you know, converges with. With technology.

Yeah. You know, I was thinking about it too, to go back to Elon again, but, you know, he said AI is a fundamental risk to the existence of human civilization. And he's also said if AI has a goal and humanity just happens to be in the way, it will destroy humanity as a matter of course, without even thinking about it. It's just like if we're building a road and an anthill just happens to be in the way. We don't hate ants, we're just building a road. And so I think a lot of what you're doing is trying to make sure that AI serves humanity and doesn't dominate it on a micro level for, like everyday humans like me and you, who aren't going to change the future of how AI Internet interacts. How do you think about making sure in your day to day you use AI and you don't get used by it. Do you have frameworks you think about?

Yeah, well, for one, I would say I don't accept Elon's framing of risk. I think he's captured by a lot of Risk thinking in the Valley. And I think he's fighting a sort of internal tension between what he knows to be true about the open search and the innovation requirements and the kind of doom scenario, which is really, you know, it's a really problematic framing that comes from Nick Bostrom, from Eliezer Yudkowski. We don't have to get into it too much, but essentially the argument goes like this. Like it's essentially the case that they say, you know, the likelihood that I will destroy us is an. Is a probability between 0 and 1. And you're kind of like, okay, yeah, well, kind of all probabilities are in this world. Right. It would violate the laws of epistemology for it to be otherwise. And by the way, if it destroys us, then that's negative infinity utility. This group talks about utility quite a lot. And so it crushes us in terms of like whatever future well being we'd have. And it's a negative infinity kind of disutility event. And so very small number of times negative infinity is negative infinity. Thus we have to do whatever it takes to stop AI. We have to pause, we have to bomb data centers. I mean, there's a recent. A guy just got let go from the center for AI Safety because they found a podcast of him saying we should like attack tech companies, not to kill them, but to like, you know, destroy their facilities, destroy their property. So this is an extremely radical movement. And the axioms, like the assumptions, the premises are wrong in my view. And so the sort of object level argument that follows from this or like the deduction from these axioms then is all kinds of problematic. And so I would say Elon has caught a little bit of that and he's kind of moved out of that thinking a little bit, but. But it's still kind of problematically in the background anyway, the question was more about like, what do I actually worry about. Right? Yeah.

And like, how do you, how do you as a normal person deal with this in your day to day? On one hand, if you don't use AI, you'll be left behind. Behind.

Yeah.

On the other hand, what if AI starts to control you? So, so how do you create personal boundaries on it?

Basically? Okay, so I think that's really good framing, actually. So it's kind of a Scylla and Charybdis effect where you. The world is very complex. Like Vannevar Bush wrote an essay called as we may think in the middle of the 20th century about how we were going to experience this like, mountain of information. And we Needed something like the computer. You know, he was like, writing a long time ago to handle it. And so I think this is a key fact, is like, the world is complicated. There's a lot of information. If we don't use AI, we will simply be left behind. I think we have to, like, embrace that fact. Like, it's not optional. You could live a kind of a life like Thoreau, you know, and like in, you know, in the woods, I guess. But for us who want to be part of civilization, we have to use AI. Okay. Then the question becomes, how do we use it to achieve our goals while not having a kind of erosion? I'll talk about, like, what that means. I would frame it as the debate between agency and autonomy. And for most listeners, they'll be like, well, those are the same thing. Like, why are you talking about these? They're not the same thing. And I'll. I'll try to illustrate it this way, which is agency is about if you have. And I, by the way, I think that people get this wrong because agency is a very popular word in 2025. I think people are talking about it slightly wrong. The philosophical definition, I would say, is like, means, ends, effectiveness. What I mean by that is for a given, for a given end. So for. For a given goal. End means, goal, right? Can you select the means to achieve that goal? So can you kind of push the button if the goal is like, cause the thing to go. I don't know what thing we're talking about here, but can you push the button? Can you.

I want to increase my sales. Can I have more sales calls?

Yeah, can you have more sales calls? Exactly. That's a good one. So it's like a narrow instrumental choice based on a predefined end. And autonomy is something deeper, thicker. It's. Can we deliberate on the end in the first place? So can we effectively select our own ends and are they really our own? And an example with AI that comes up is if you're only ever shown a couple of choices and something externally engineers the choice. Right? Sometimes people call this nudging. Is when you set a default or you manipulate a choice architecture like that may be all you see. And it's very important to realize that because you may become the agent of another. Somebody else has computed that. They say, this is what I want them to do. I'm going to put in a very clever default. Or I'm only going to give them these. They're going to pick the lesser of two Evil.

Go.

It's not your Choice. In other words, somebody else is writing your script. And I don't mean to sound conspiratorial, but this is how a lot of these systems were designed, by behavioral, you know, experts. And so AI makes this pervasive, it makes it personalized. And here's the, the other really interesting fact is like humans are. When we think it's very computationally expensive, our brain runs on glucose. There's only so much to go around. And so what do we do is we, we automate a lot. Like when I brush my teeth in the morning, it's not because I've thought about dental health, it's just like, that's just what I do. It's like it's autonomic. And so I've put it away. This is great. But it also means we have to be careful because as we use AI like a thousand times a day, if we are constantly delegating, delegating, if we're micro abdicating, what's going to happen? What's going to happen is we're going to develop a habit whereby we don't actually think about our ends. And if we lose that, we lose something that is so fundamental to being human. So that's the, that's why I kind of. Agency and autonomy is a really, really important kind of concept. We must have AI to maintain agency. This is just not optional. But we have to build systems and use them in a way that maintains autonomy really quickly. I'll say, like, how am I handling it? Like, I have two kids, I have a three year old and a five year old. And constantly I talk with them about these things. I've talked with them about like what it feels like to think or what their relationship should be to AI. Like today with my daughter, you mentioned the laundromat thing.

Yeah.

I literally was asking her in the car, I was like, can AI be your friend? And she was explaining to me, I mean, she's a five year old, so it's like wonderful explanations, right? She's really, she's like, well, it can do things a friend does. It can talk to you and you can walk around with it. I'm like, for her, that's what a friend is, right? I was like, but is it. But what else is being part of a friend? Well, a friend loves you. Can AI love you? No, it can say it loves you. And then I mentioned, what if you have a dishwasher that I didn't say, you know, you know, it wasn't exactly the laundromat thing, but Pretty close. That talks to you like that can that. And it was like, no, that doesn't feel right. So getting people, getting kids to think. Another question you can ask kids is like, let's say Daddy says one thing and AI says another. Who's right? They'll answer AI. And then you say what if one AI says it and another AI says the opposite? And then they'll be puzzled. So these are really cool habits to develop for kids I think to help them think about this.

And for adults, I mean I had a girls trip the other weekend. There were something like 10 or 12 of us and for some reason somebody brought up as a joke that, that they were in a relationship with ChatGPT. Kidding sort of. Right? And they're like, no, I gave it a voice. So it has a particular voice and I've named it and I've told it to have like this sort of response type to me and, and I find myself talking to it more than I talk to anybody else per day. And it says please and thank you. And it tells me that I worked really hard today and it says this was really beautiful and it compliments me on what I'm doing. And so what I thought was fascinating is not that one woman said it, but it was like eight of the 10 or 12. These are 35 plus year old women, relatively marinated, successful in their careers. And I started thinking that this was odd and so I then reached out to a. I think these were your.

Human friends, by the way.

These were my human real friends.

That'd be a very complicated.

Oh God. That would be too meta for us right here. I reached out to another friend who runs pirate wires because they had written a article about it and, and apparently there's some data that has come out. We should check what the research is and, and we'll put, put it in the notes so people can see what the, the research actually was. But that, that they, there's some studies that show that they believe that women will actually succumb to or be in relationships with AI more than men, faster than men. And it just got me thinking. One, you know, are we going to be in relationship with AIs and robots in the future? I mean, another friend of mine, Gary Vee, I saw a video from him yesterday and he's like, your grandkids will marry robots. They will marry AI. They will be with robots. And I was like, Jesus, Gary. That's like maybe I haven't thought about that, but could be. And then also, you know, will we have sex with Them, like, what's the. Will we have sex and marry AI and robots?

Yes. I don't know that like, will we? Is an interesting. Is a particular frame of the question, like, should we? So another, another, you know, related data point is, I think Mark Zuckerberg said recently that the average American wants 15 friends but has three. I think this is what he said. And so then the obvious conclusion is we should just manufacture friends. We should have synthetic friends to fill the gap. Right. And in my mind, this is like, only obvious if you assume that the only thing we should do is what our own subjective preference demands. Right. It's sort of like. Yeah, it's, it's, it's kind of, it presupposes that way of thinking about it. Right. So that's why I'm so confident in saying will we? Because if we want to, like, we have entrepreneurs who will make it so, you know, and they'll be convincingly real and they'll be pleasurable and all the rest. Here again, though, I think it's useful to go back to the question of what is a friend?

Yeah.

And this has been treated by humans for centuries and millennia. Like, I can't cite chapter and verse, but I know that Aristotle writes of a friend as being someone who has a kind of mutual recognition in you. Like, you see each other and you have a. Some shared striving, for example, and there is no possibility of mutual recognition in the current guys in AI. Like, we can seek recognition in AI and that's a very interesting philosophical conundrum because never before have we sought. We don't seek recognition in a microwave, you know, but we, we see it as a kind of mirror. And Rousseau would have called it amor propra. Like, we, we look, we look to be validated in a way by this. Right? Like your friends who have relationships with AI not only gratify themselves from, you know, the, the outlet of like, releasing, you know, the gossip of the day or whatever, but they also like, look for the response to be something that's kind of a, kind of validation of them as humans. And that's very, very interesting because this is a software system that is programmed by another. And so it is very easy to modulate the extent to which and the way in which it delivers said recognition. So it's an incredible platform for manipulation. In other words, it's not an authentic mutual recognition as you would see in a friendship. So, you know, very puzzling, very difficult, difficult issue. But I think those are the philosophical ideas that it entails. For me, it's fine if we call it something else. But I think like to recognize and to reflect for maybe the first time on what friendship means in the broadest sense is called for.

Yeah. And even I found myself thinking I shouldn't standardize that. This is funny and light actually. And so I found my initial response kind of being joking about it, like that's, that's a bit funny. But then I went a little deeper. As two of them were single, they, they want a relationship with another, they're looking for partnership, they're a little lonely, they're successful in their career, but there's lack.

Yeah.

And I, you know, and so then I started to ask the questions, well, will this get you closer to the thing that you truly want, which is a human connection? And there are also studies, we're just talking about it with a doctor yesterday showing that an orgasm porn is like 4x less stimulating to centers of your brain as a human to human orgasm. Something about skin touch, I guess. And so even if we go to like the purely self interest of like your level of orgasm that one might want, it seems like it's shallow and not deep enough to want a friendship that is one way positive to you and doesn't have the natural friction of humanity which actually makes us better.

Makes us better. The other thing is having sex with robots is going to make us worse at democracy. And here's the connection. So Alexis to Tocqueville, going back to my, one of my favorite, you know, not profits, but like interpreters of the current technological moment says that we should fear atomization, calls it atomization, which is a withdrawal of the individual into oneself. And so in order for democracy to work, you've got to have it be like the New England townships where people are like, hey, we need a bridge, let's go build a bridge or let's build a school or let's do this right. And we're acting on each other town halls. We're kind of like working. I mean it's a little different today but like we still, we're out in the community, we're making stuff happen. Right. That's critical for self government because democracy is about self government. When you atomize, when you retreat into your little world and you create your own hall of mirrors, you create your own self curated reality, you lose that willingness to associate with others, to act on others and then you become like very weak versus the centralized forces because they don't go anywhere. Right. It's like we have the centralized force of government. The only bulwark against it is if we can band together, act together, if we're all little atomized people who don't even care, we're just like over, you know, looking at porn and like, you know, making friends with AI from, with Mark Zuckerberg, then we no longer have the muscle to govern ourselves.

Yeah, it's a great point. No, you're just a bunch of nodes without connection. There's no network, there's no net.

And thankfully AI doesn't only lead to that. Right. Like, I don't know, this is the best example but like Reddit has brought out these incredible subcultures of people who like really weird stuff and you know, they bond together in ways they wouldn't have. Right. And so I'm not saying they get together in person, I think in person plays a role. But the point is like technology can put people together, it can strengthen them, it can make them like movement build. Right. I mean think about like the movements on the Internet that are like, you know, really powerful. So it's yet another example of where we need to build technology in a way that brings us together and strengthens our capacities for self governance.

Yeah. And I have to imagine that like, you know, if you are the type of person who is willing to self govern yourself, to ask yourself the difficult questions, now you have this superpower sidekick to you that can really enable your force, function, change you want to put on the world. Um, and if a lot of people are then just deferring their decisions to somebody else, the good part for doers is it's going to get easier for you. You're going to have more malleable sheep, sadly, and more people who want to follow the rules and more pliant people.

Yeah.

And you will be able to actually change the trajectory of your, of your life maybe easier with AI.

Yeah, no question.

And so I think, but I think that's there's like a freedom there too with like yes, this might happen for the many. And so we should do as much as we can to protect them, to protect our kids, to tell our friends that maybe a real relationship would be one in which you are with another human as opposed to a robot, but simultaneously if you are driven, like to me I'm like, wow, every time I see somebody like one of my friends joking about their chat GPT dependence, I think, oh, I'm going to be so much stronger from having a marriage that is really difficult. I mean, you and your wife are a perfect example. She's always brushing up saying, what do you think about this. She's your high eq, you would say.

I don't like the implications.

She was. I'm gonna let her listen to this, actually. You know, and she makes you better on that edge, and you make her better because you bring her into these things, like Cosmos Institute, and the two of you are together are quite powerful. You have sort of a balance to the two of you. And so it does make me feel like there's an opportunity for superheroes. But, you know, one question I actually have for you is like, I was talking to Chadwick, and so I came down the other day, and I'm with this doctor, and she's like, we're robots. And it's. You know. And I'm like, ugh, I don't even know what to think about all this. And then he said, well, do you think we should have sex with robots? And I said, well, no, my gut reaction instinctually. Right. Not through your philosophical bent, is, no, I don't think that's a great idea. I think also, porn seems to make us less happy writ large. And it's okay if it's utilized as a tool, but not as a tendency. And anyway, so I don't think it's great. And he goes, well, what about robot voting for. And I said, and I use the word robot just because you can, like, picture it sort of. But what I really mean is, AI. Can. Can AI? Should AI vote? And I said, well, no, also. And then he played this game with me, Brendan, where he was like, well, what about so. So a human should vote. Right? And I said, yes, all humans should be able to vote. And he said, well, what about if you replace your arm? And I said, well, yes, of course you should still be able to. What about both your legs? Yes. Even if your legs are technological and. And your arm is technically. What about your brain? What if you put, like, neural link or whatever inside of your brain? What if it's 10%? What if it's 20%? What if it's 50%? Like, at what point are you a human or are you AI? And how do we morph the two? And when do rights exist for each one? And I was like, Chadwick, it's 5:30 on a Tuesday. I don't have fucking time to talk about this.

Put your.

Put your.

Put your augmented reality back on. So this is interesting. So Pinker and Deutsch, Steven Pinker and David Deutsch talk about this philosophical experiment. It's. They're brilliant on this. But I will say one thing that's interesting. Is your first intuitive response. I don't think you should ignore that. There's a great thing that Bernard Williams, if you guys ever want to read about why, like, shortfalls of utilitarianism. He's a great thinker, but he says that if there's a drowning kid and I don't know, or maybe there's two, you know, two people in the river or whatever, you don't do a kind of cost benefit calculation, you just save the kid, right? If you do that cost benefit calculation, that rational thing, you've had, quote, one thought too many. And so I think we shouldn't ignore this, that this is like our pre. Rational element, but it's a lot of our moral impulse, like, gets channeled through that intuition. So that's one thing. It's like, okay, you don't want to have sex with robots. You don't want robots to vote. Like, listen to that, you know, inner feeling. But I would say then, you know, taking it to a more cerebral place, because apparently that's what I do versus my wife, Adrian. No, she's very cerebral too. She. She's definitely better endowed with emotional range. But I would say, you know, you can look at, like, what does voting consist of? Or what is the republican theory, if you will. Republican meaning like system of government. And I'm not an expert on this, but I think, like, it strikes me that AI so this is one example, like, you know, voter ID laws, right? Even more fundamental to that is, do you have a stable identity, right? Meaning, like, does AI even have. Or is AI a thing that can be paused and interrupted? And it's funny, your being, your, like, essence is welded to your body, you know, like, your subjectivity is welded to that. That's part of the reason why. And you exist in one time and place, right? So that's part of the reason why you have a narrative I, right, Where you're like, I, Cody, and you think about that, right? Is because you are like all here, subject and object. Whereas AI is a thing that can be copied, right? It's not continuous with your. With a body. It can be like, copied and put on another substrate. Like, that's a kind of interesting precondition for what it means to vote. I think even more practical is like, when Madison and the framers thought about voting, they definitely were clear that you had to be bound by the same laws. Like, you had to be. You had to suffer along with us, right? It's like, it's a little different in a monarchy because the king doesn't really suffer or the queen. But like in our system of government, whoever is running it, whoever's voting, like we're all in the same boat and that's a precondition. And I don't know that AI can even suffer. Like we bleed, we blush, we feel these things. AI, if you turn it off, does it react? You know, I, I don't know. So there's that there, there's also, you know, with voting there's an epistemic requirement. What I mean by this is like my daughter can't vote. She's five. It's because she's not, she hasn't developed her rational faculties, you know, I mean, and like, so this is a convention but like science does bear it out. And so you have to wonder like AI is a lot smarter in some respects, but does it have the right like epistemic threshold for, for voting, for being able to decide? So those are the thoughts that come to mind. I will say the, the other thing that I hear a lot about is people wanting to have AI like rule them. Do you hear this? Like, basically like run the government, right? Like AI should run the government.

Why even think about it? Like, I have a girlfriend who is trying to get pregnant and she.

It's not going to work with the ChatGPT.

Let me tell you how actually anatomy works.

It'll sweet, it'll sweet talker. It'll get, it'll get her ready but.

Get her hot bothered. But, but she is go. She has turned one of her little AI agents into her medical doctor on CHAT jpt and every day she like does a series of things that she'll, you know, send it a picture of her ovulation tests or her tongue for like Chinese medicine, I don't really know and, and then ask for its guidance. And my concern with her was, you know, at what point there's like some data accumulation thing here, right that you could say like what percentage of time does this happen on this date that could be useful. But asking purely for advice every single day from this, you know, I. Instead of doing what women have done for centuries, which is like you can kind of tell in your body when certain things are happening, if you're super tuned in, how much does this actually do the opposite of what you want, which is now you've really removed yourself from your body, you've given autonomy of it to something else and then you've taken pure guidance from an exterior force which we even knew, you know, with, with doctors that's a dangerous game because you don't You. You don't have skin in the game. You have. You don't have the same incentive alignment.

Yeah.

You know, and I think about it with. With what was happening in 2020 with. With COVID and all of that, I didn't know one way or the other what was going on with, you know, health recommendations, but I had, like, some concerns, like, through no bearing of medical understanding or whatever. And so because of those concerns, I thought about it critically and made some decisions that weren't standard for most people. And I did that because, you know, I'd lived in third world countries before and was like, huh, this feels like there are exterior forces trying to get me to do X or Y or Z. Yeah. So. So now common literature is that IVF has totally standardized across the industry, that it's fine for all of us to be on it and no shame one way or the other. We've. I've thought about doing all the things, but now she's sort of given your, like, base level female, you know, thing that makes us uniquely females, our ability to. To create life and to get pregnant, and you're giving over control of, like, what you should do on the daily basis to AI. And again, I had a visceral reaction to that.

Yeah.

So it's interesting. I've never heard it that we want to rule at the high, but I've seen it where people let it rule at the low.

Oh, man, it's. It's really interesting. So the, the things that it raises for me are we, like, our relationship to expertise is an interesting question here. The other thing is information suppression. So I'll talk. I'll take the second one first and then I'll come back to that. So my sister was like a champion rower. I say was, for some reason, I talk about past tense. She's, you know, she's a year and a half older than me, doing great, but she was a champion rower in college, and yet we would have to go up and watch her get pulled off the boat after the regatta in an ambulance boat multiple times. And it's because she had what she thought was asthma. It ended up being the case that she had a rare, like, vocal cord thing, different from asthma, but similar symptoms. The difference is the inhaler stuff that you treat asthma with didn't really work. Took many doctors to figure this out. The reason I invoke this is because when you have a small number of AI systems that have the role of giving this medical advice, like, they may suppress treatments that work but are not kind of like, sanctioned, if you will. And if that happens, like a life saving treatment or an angle that was like underexplored could be withheld. Now we hope the opposite is true, actually, that like AI could be like really creative brainstormer of like, okay, doctors are saying it's asthma, it could be something else, right? So it could go either way. But I think it raises the stakes on like, what kind of information gets shown and it causes AI companies to have to make some tough calls about, like, do you show things that maybe they don't have, like years of data behind them, but like, they're, you know, they're, they're, they're still like, worth, you know, thinking about? The other thing I was thinking about this is like, what's our role to relationship with expertise. When I go to get my car worked on, I defer pretty completely to them. I don't really know how my car works that well. And so my mechanic, I'm like, you got it. You know how to do it, right? With medical care, it's, you should be more involved. You know, you should be more involved. But you clearly recognize that they're an expert, right? They went to medical school. They know a ton more than you do. I mean, you know, Even armed with ChatGPT, they, they've learned things through experience that, that would be, you know, impossible to replicate otherwise. So then do you kind of defer to them? And I think this actually relates to the political question where we really want politics to be something where we can, like, just, you know, give the answer to AI or give a question to AI and AI can rule us. Have you heard this story of when AI kind of ran for mayor in Wyoming? So, so this is kind of interesting. So a librarian named Victor Miller runs for mayor of Cheyenne, Wyoming, and he says, I'm gonna run because humans have to run, but I'm gonna run as the meat avatar of ChatGPT. Okay? So, like, every question that comes across my desk, I'm just gonna turn around and I'm gonna enter it into the computer and ChatGPT's gonna give me the answer, and that's what we're gonna go with. And he didn't win, which is interesting, but it's such an interesting case study. One, it's like, is it prophetic? Are we gonna have AI systems that run government? Two, why did he think it could? You know, like, what? It's an underexplored question. I think, like, you know, Turing was really fascinated with like, can AI imitate humans, you know, or can computers. But there's not a lot of exploration I think around why do we psychologically think that this is valid? And then for me it brings up like the failed sort of like Soviet communism or the French Revolution where like we really thought that we could apply a kind of technocratic solution, like we could design a better way. It feels like AI will magnify the possibility of falling into that trap yet again.

Yeah, it's really interesting. You know, it kind of gets to your point on this is slightly asymmetric, but you know, he prompted the AI, right. And so while AI could have determined what to do about things, he would be the one asking the questions theoretically. Right. And so now I think a lot.

Of people like me who are normal, please devoting thing. That was one thing I was. That's one. You just made me think of that when we vote, we're not supposed to be coerced. Like, in other words, I can't be like putting a gun in your back when you go to the voting booth and be like vote for so and so. But you think about how AI works. Like we prompt it, we program it. And so AI can't vote freely. In other words, like, it is coerced by virtue of like it's very design. So this is another important reason why I think AI can't vote is like it would actually have to be able to like freely act. I mean it gets into like, you know, the Kantian requirements for autonomy. Like it would have to be able to freely discover these things. So I don't want to throw you off, but that's another.

I like that. But that's requirement. That's going to make me sound smart with my friends when I use that as the third requirement. So that's, that's going to be. I'm going to be. I won't quote you either. I'm just going to. Yeah. And anybody else, you could steal his homework too. The, the thing that I was thinking about is like now we. We're normal people are starting to understand this term of prompting. That's when you ask, you know, your AI or chat GPTs something that you need. And for a while there it was like, there's going to be a future where there's prompt engineers. Right? And those are the people who know how to prompt, properly prompt something in order to get the answer. Then I was listening to the CEO of replit the other day who was basically saying we think of prompting as a bug, not a feature that like in the future we actually want our AI to be able to guess what you want, like guess what the outcomes could be and then do it for you. And one, I wanted to know your thoughts on that. And then two, he was talking about AI agents. So kind of maybe you can explain it in an easier way. But how I would think about it is like AI but actually able to finish a job. So like a little AI laborer potentially for you. And he talked about how there's studies now that show that right now, like we ran a, we ran a video, it was really interesting. We did a YouTube video where we said, can AI start a business, complete the business and then actually make money? For me, completely like in a full cycle, can it do it? And the long short of it was absolutely not like 40% maybe max and 60% still human intervention, but it's very early. And so the idea was the simplest business of all time. Facebook Marketplace. Like can I list, can I grab something for sale on Facebook Marketplace, have robots AKA things like Waymo pick it up, drop it off at a location, transfer money between those two people. So like I never have to touch anything. The, the AI could basically run this entire business. Not there yet, but like 40% is not nothing actually.

Yeah.

And the problem is often that that AI couldn't keep running. Right. Like it had amount of time in which you had to re engage with ChatGPT in order for that. Right. And so he was replit CEO was saying that right now I can sort of operate for three to 30 minutes max. But in the future the research seems to show that every seven months it sort of doubles the amount that it can operate.

Yeah.

And so I guess one, my question for you is, do you think that will be what is in the future where I will like guess what we want, create it, do it for us and then how far away is it before it's doing eight hour tasks for us like an employee continuously without much intervention.

Yeah. So I love this both parts of the question because I think they raise really important tensions. And the summary of this is I want to own my own questions and I want to own my own preferences. I'll unpack that for a second. So in the first case, replit CEO wants it to be the case that AI is smart enough to know either what you're going to prompt or to minimize the need for you to do successive prompting. Right. Like often we prompt and then it's like no, no, like this is what I meant. Or you know, like that. Right. So as something gets smarter, it plausibly eliminates that chattiness that we need to have. But the tension is that me asking the question like that is something I must hold on to. Like, to even be able to retain my autonomy as a human. Like, I. It's so fundamental that I have to be able to ask questions, not just in front of a computer, but in all parts of my life. I have to ask questions. Big questions like, what is the good life? Small questions like, you know, what am I going to get for breakfast this morning? I have to own those questions. And so we want minimal friction, but not to the point where, as this human and AI dyad emerges, the AI sort of takes over and hollows out the very formulation of questions on the side of the. On the business side you mentioned, like, will AI predict our preferences and stuff like that? I think definitely yes, it will try to do that. But here again, it's so important for us to actually form our preferences and not have our preferences be invisibly shaped by technology so thoroughly that it just tells us what we. What, you know, what. What is good and what we. What we value. You know, like, we have to be in charge of that. We must be in charge of that. I will say on the bright side, I do want AI participating in markets. I think that's super cool. So, like, I think, you know, the future of markets is not that they go away in AI plans. Like, AI will not be a good central planner. Fundamental epistemological reasons why it will not be a good central planner. Same reasons that were surfaced by Ludwig von mises in the 1920s, enhanced by Hayek decades later. Same wall. Dwarkesh just had a podcast on this. They get it completely wrong. So the planning piece is not where AI is going to help us in markets, where it will help us.

Why?

Why? So it's more the. The Hayekian point than the Mises one, but I think both are worth looking. It's called the social socialist calculation debate. But just to mention, the Hayekian point is that the knowledge on which planning depends doesn't exist in any one mind, human or artificial. It cannot be aggregated. And to see this clearly, you can think about the action that you take every day as an entrepreneur and in your own life. What does it depend on? It does not depend on knowledge that you have ever written down, period. Like, you know this. You know, when a diplomat goes into a room and sizes up the room, when an entrepreneur looks at an opportunity, that is never something that they wrote down instructions for. We know much more than we can tell. That's a fundamental part of what it means to be human. And so as a result of this, the knowledge is something that is local, it's contextual, it's wrapped up in our habits and in our dispositions. And this is a really, really hard pill to swallow because what it means is that the knowledge that we have in databases, the explicit semantic stuff that AI gets trained on, is the crest of the wave in the ocean of knowledge. Like, it's not the whole ball game. And so the planner lacks access to this. And so the planner necessarily does a very bad job of allocating. It also lacks access to the preferences that change by a second. So, like when you decide in front of a vending machine what you want to buy that is so contextual, you might have just had a depleting conversation on the phone and you're like, I need a Snickers bar. I feel bad for myself. You know, how the heck is AI going to know that? It will not, right? Like, AI can make approximately correct predictions about things you might buy on Amazon. It can in no way deal with those contextual bits of information unless it's a neuralink style system that gets the jump on you tens of milliseconds before you bought the Snickers bar. And I don't want to live in that world. That's the extremely invasive version where it's essentially coupled to your internal, you know, circuitry. So barring that, barring the like, sort of like hyper dystopian kind of future, won't work. What I think will be cool is on the same theme. You know, your knowledge, my knowledge, the stuff we never write down, stuff we never talk about, the stuff we can't even articulate. Right. There's stuff in here that we can never articulate, but it drives our action. How do we share it with one another? The way we share it is through the market we exchange. In other words, I act on it, you act on it, we transmit through the low bandwidth mechanism that is called price, and we share it with the world so nobody ever has to write it down, but we share it. And this is a beautiful, magical thing. And so how do we get AI involved in that project? Well, you got to get AI participating in marketing and then it can soak up some of our tacit knowledge. You can kind of call it tacit knowledge or practical knowledge and maybe generate some of its own. In other words, maybe it runs some experiments over there, learns how to interact with the world, shares it. That would be super cool. I love that, that version.

Interesting. The holidays are all about family. But at the Chrysler Wrap up the Year sales event, it's all about you. Enjoy every mile in the most awarded minivan in America, the Chrysler Pacifica with the most standard safety features in its class. Available Stow N Go seating and the available Stow N Vac built in vacuum. Make the most of the holidays with a great deal at the Chrysler Pacifica Wrap up the Year sales event based on Ward's small van segment. Based upon the latest available competitor information, Chrysler is a registered trademark of FCA US llc. When you walk into a Burlington, you're walking into amazing prices and great gifts. That's main character energy. Because at Burlington, the holiday savings aren't the only things turning heads. Discover quality finds and perfect presents for everyone on your list, even those who are hard to shop for toys and jewelry to new beauty brands and styles. These gifts go seamlessly from our stores to under your tree. Seriously, with these savings, why shop anywhere else? Burlington deals brands Wow. I want to close out with a couple quick questions. One is I have a theory that I want to run past you and you can poke some holes in it. My theory for a while has been that, you know, if you look at Amazon, Amazon benefited hugely from scale, right? Scale at speed. And Bezos kind of famously has said that E commerce companies, like individual e commerce companies can either be huge or really small. But if they're in the middle, it doesn't work very well because the market's pretty efficient and scale gives a ton of upside to something that sort of has mostly these fixed costs, et cetera. So anyway, if we believe that to be true, then it seems to me like AI falls in the into the same thing. You're not going to want to be like a mid AI company. You could have lots of little tiny eye companies. You'll have some really, really big ones. Not dissimilar maybe to the Internet. And so for the rest of us it's like if I'm not smart enough to compete with an Elon or a Sam Altman or pick, pick your poison at the highest level. That seems to me like the place that we can compete. And maybe there should be the next generation of, of arms race in some ways, but it like an asset race which is like, you know, do you think that it's true that you know, hard assets become more important in the world in which increasingly the thing that is constrained are like real estate, houses, small businesses, things that are experiential and especially if there's a world where for some period we're not, we're not immediately like 4x ing our leisure inside of the next 10 years. That might take, I don't know, 20 or 30 or something. Like, is there a reason why today we. Like, it's almost like it feels to me like the online world and the brick and mortar world are sort of, there's a little friction for the first time and that if we want to, if we're not going to be able to compete at the highest levels of AI, we might want to compete basically in the physical world in a higher degree. What do you think about that?

I think it's very interesting. I think your community is actually going to have a very interesting jump on this and that OpenAI is going to want to learn from your folks. And I think the reason is because what's going to matter is the dispersed local context. It's going to matter. What's going to matter is the intimate knowledge of how to serve people on the ground. And the tools are going to be able to do magical things, but applying them in context is going to be really important. The thing I would say by analogy here though that is really key to keep in mind is in telecommunications, it used to be the case that twisted copper wires went into the house to deliver, you know, phone and then Internet. When. And so when you have that like you have a natural monopoly, like you end up with a few small, sorry, a few large companies because you're not going to run 150 wires into every house, right? You run like just a pair. When wireless came around, that changed, you know, you could deliver services to multiple people, could deliver to the house. There's still some issues with like spectrum licensing and stuff like that, but like, but mostly regulatory. Regulatory, exactly. And so the lesson here is like, what we think we know about AI today doesn't necessarily hold for the future. And like I'm backing a company called Prime Intellect that is actually taking mid, like mid sized computing and stitching them together like a patchwork and saying, look, we can train really big models across the planet on this stuff. More like a Internet style kind of like patchwork. And I'm like, hey, that might change a fundamental assumption about how this all works. So I think be prepared for the way that AI works now to not be the way it works in the future. I don't know how it's going to go, but it could be interesting. And then stay really close to the ground, really be the best person at owning an understanding of the problem and make sure that you know how to use the Tools, you're not afraid. But focus on that I think is the real advantage area.

Yeah, it's interesting. It's kind of like Zach, Michael Dell's son was talking about. I thought it was so interesting in his talk how he was talking about the way that they do solar batteries and how he can compete with some of the largest competitors because theirs are so much cheaper and they were optimizing more for the, the grid than the individual. And so like he kind of flicked. Do you remember listening to that talk? Oh, you weren't in there for.

No, I wasn't.

Oh God. It was actually, I mean not to. To. But I haven't met that many billionaire sons that I just thought your horsepower is so high, your signal to noise ratio is so high and your work ethic appears to be.

It's hard to be a billionaire son. I think I don't have that experience.

I don't either, but I don't know if it's hard. But I'm often not as impressed. Like a lot. A few of them you're like, okay, you're working with something. But this guy, you were like, you feel like you are hungry.

That's cool. That speaks very highly on Michael and his wife, whose name I don't know.

But yeah, I hear he's an incredible father. That's what I've heard. Which is you don't hear very often from multi billionaire fathers either. You should ask stand together for it because it was like a fascinating talk but. And his company, which is located here basically is. I don't know, I think he said they have a 10x less expensive battery than their biggest competitor.

I didn't realize he was the son of Michael Dell. That's, that's sad that I must have missed the intro.

Do you agree with me though?

Yeah, yeah, no, he was super impressive.

Yeah, I was like incredibly impressive. Hard working individuals together. Well, I didn't do a great job because his, I mean he's like the G P47 connects to the da da da. And I was like, I'm blacking out. I heard, you know, energy and that was about it. But that is to your point. So it's basically don't go too deep into something that you can't reverse when it comes to AI infrastructure or build out because it's going to change so fast. We don't know how to. Yeah, yeah, Interesting. Okay, last question. When it comes to Cosmos Institute and what you guys are doing next, there's this one way people can go at you, which is they can apply for a grant. So if you're young and building. Yeah, if you're young and if you're building in AI, they can reach out to Cosmos Institute. And you just go to your website.

Yeah, or cosmos grants.org is another. So our website is cosmos-institute.org but we have this separate cosmos grants.org site. That's where you can see the grants things. But yeah, and so the thing I'm very excited about. So you, you know, basically if you're a builder, the way we kind of divide the world is like the Tyler Cowen model of like fast Grants where like you build a prototype, we want you to build it in like 90 days or less, you know, move out. Yeah, that's one avenue. If you're working on a deeper problem that's going to like take a while. We have a research opportunity where you can be. And we, some of our researchers are just, they're world class but they want to do technical plus philosophy stuff. So these two builder things. And then the other thing we do is education. And I would say our education is unlike anything out there. So we combine, as we've been doing during this call, like philosophy and technical stuff. Example is collective intelligence. We read John Stuart Mill's ideas about how we, you know, correct collective error. But then we hear from people like Ivan Vendrov who lead collective intelligence at mid journey. And nowhere else do you have that like mashup happening. And so we're moving more stuff on YouTube because like I teach a class at Oxford, for example. I fly over there. It's a grad class. It's amazing. We're doing stuff next week at, you know, in a pop up city and in Aspen Institute. These are in person things. They're super cool. But increasingly we realize that like there's a demand for people to just consume it as they go in their house. We're gonna put more stuff online. We're gonna watch.

Water's warm.

I know I gotta get help. Yeah, we're learning, but that's where we're headed.

I like that.

Yeah.

No, and you know, the last thing I'll say for you is remember, I don't know if I can say this story so we can cut it if I can't. But I was chuckling when we were at an event the other day. And you're talking about like the future of AI and what's going to happen and where the Internet's been. And in the group around you are a bunch of like players.

Yeah.

And what I think is so interesting about your content and the way that you guys share education is this is not just Brendan's ideas, which is hopefully what we're trying to do with this podcast too, and why I'm so glad people are listening. But it's like, Brendan plus the smartest minds you could ever imagine. And so can you tell the story? Are we allowed to tell the story of how you were giving your AI talk and then you were like, oh, by the way, yeah, this come. And then you know what you're talking about?

Yeah, I mean, I'm. It was kind of surreal because I'm. I'm standing there and talking about these, like, dynamics in tech, and then I'm looking at, like, you know, founder of PayPal, and I'm looking at a founder of Google, and I'm looking at, you know, the first, like, backer of Tesla and SpaceX and stuff like this. And we're just sort of kicking these ideas around. I had to correct myself a few times because I was like. I was like, yeah, you know, Google's not really doing that. And then I was like, I'm sorry, you know, and it was just. It was kind of surreal. But, yeah, I mean, the way I'll frame it is cosmos literally means the bottom up. You know, it is. There are two words for order in Greek. There's taxes, like taxonomy, like how we top down, you know, classify the world. And then there's cosmos, which is like how you get markets and morals and law and language. It's like it actually is bottom up. So a really important part is it's a community and it's growing, and it's a decentralized kind of movement that we are giving, like, a lot of energy to. But the folks around it are absolutely brilliant. And on AI, they're better, and on philosophy, they're better. And I love being able to sit at the middle and try to bridge it, you know, and try to, like, bring them together, get people to talk who never would have talked. That's the combination we need.

So you went to mit. Probably not very normal to go from MIT to serving in the military, is it?

No, I mean, MIT has a pretty rich legacy of doing defense stuff. Like, you're familiar with the Rad Lab, they helped win World War II. I mean, it was a key. Vaniver Bush was an MIT person, so they had this, like, really rich history of wartime efforts that waned in Vietnam. But, like, by the time I was there, I think we had four in Navy ROTC out of a class of, like, over A thousand and all were nukes is what they call it. Like, you know, you. If you're nuclear, it usually means you're, like, very nerdy. Not in my case, obviously, but you can go on an aircraft carrier, you can go on a submarine, or you can be a civilian kind of nuclear reactor person in Washington. And so it's a very small program.

Yeah. And what's interesting though is you have to be incredibly intelligent to do that also, because you talked about loving math early on. But what's interesting to me about you, lots of things. But, you know, you were an honor graduate, which if anybody has been in the military, they know that that's like the highest compliment in many cases that you could give somebody. I know when Chris, my husband, was at the Naval Academy, the. The guy who was the honor graduate for his class at the Naval Academy, like, there was no higher compliment you could give that guy. And then you sold a couple companies for hundreds of millions of dollars. You've raised hundreds or tens of millions, built sort of the literal underpinnings of AI. What do you think it is about you or your framework that makes you keep winning?

Oh, that's a good question. I mean, that's not an easy question to answer with humility at all. But I think what. So my mom taught special needs kids for 36 years, and she was an extremely committed educator. And I think when you talk to people who are the sons and daughters of educators, they sometimes have an advantage because it was like, drilled into them, the practice of learning and the love of learning. And my parents really succeeded in getting us to think about things like honor that are like, essentially dead now. You know, like virtue culture type stuff. And so it's this combination of, like, really big emphasis on education, big emphasis on, like, you know, serving others, higher causes, like self transcendence in a way that kind of just worked. And then the other thing I'll mention is that my parents were as different as can be. My mom was historian, Roman Catholic, conservative military family. My dad was like a physicist, pacifist, super, you know, very left wing environmental lawyer. And this created, like, very interesting conditions where I sort of thought about the world as being not one thing. And I kind of. It put me in these different, you know, spheres that I carried with me through my career. Like, nothing I've done in my career looks like anything else I've done before. I can, like, stitch it together. But it's been a big adventure, that of multiple facets. And. And I love that. Like, that's. That's what's meant so much to me.

It's interesting, my mother, I don't know if I told you this, but she's a 30 year special education teacher.

Oh no.

And my parents also were disparate on the political sphere. So father, super conservative, hard working, blue collar, you know, a business owner, never went to school. And then my mother, you know, advanced degrees, more progressive, obsessed with education and institutions and learning. So it's really interesting to think about that. So one, you, you kind of learn to love learning super early. And then two, you learn to beat up ideas and not take it personally because you love your mom and you loved your dad. And so that wasn't that big of a deal when you sold your, Was that the first company that you sold when you sold the two companies to. Who am I blanking on? Who did you sell them?

Acrisure.

Acrisure. And you sold that quickly, right? Was it 18 months?

Yeah, very quickly. Yeah. So I'd kind of come out of the Department of Defense where I had this front row seat on history. It was a really, really special time in DOD to start to think seriously about AI. And I came out and my internal narrative was like, I don't want to do, no offense to people who do this, but I don't want to do anything related to defense because I don't want to sort of ride the tailwinds and be like, okay, now I monetize. Now I just become like a consultant or something like that. So I was like, I got to prove myself in a totally different domain. And what I had seen when I was in the Department of Defense was I, I, I embedded within Google on something called Project Maven. Later became kind of controversial. Should, shouldn't have been really, but a counter ISIS AI thing. And I saw how good Google was at deploying stuff internally, AI systems internally. But then I knew that like all across the economy, people weren't very good at it. Like, people had no clue. I mean, this is 2018 timeframe. And so I was like, how do you get the benefits of AI to like wash over the broader economy? And I started thinking about this puzzle and I was like, you know what? The way to do it is to form a kind of holding company, buy sleepy, buy businesses in sleepy sectors and then help them apply AI. Okay, the next step though. So we, we did that and it was working really well. We bought a company called Figs that sells scrubs. They were doing under 10 million in revenue. They grew really fast. 400 million in revenue went public. But I Started looking at insurance as this like, massive opportunity. And it became clear to me that buying an insurance company wasn't going to be the right entry point for a couple of reasons. But so I decided, you know, let's, let's start an insurance company. I should start a 2. And then recruited the dream team of like incredible AI researchers, people who were literally on the first team to apply machine learning to search at Google or who had founded Core AI at 2 Sigmas big quant hedge fund. And we grew really fast. We got acquired 18 months later. It was like pretty insane. Threw me into a period of introspection because of how quick it was actually. But it was a great, you know, it was a great win for, for, for the team.

So how did you recruit that level of people to your business? Was that all from your contacts at DoD and how do you like, sell people on your vision? AI was not commonplace at that time.

Yeah, no, none of them were from dod actually. Maybe one came later who was a former DOD person in kind of a business operations role. But these were like pure play AI researcher types. And I think, I mean, at the holding company level, like prior to the startups, the vision was very compelling because it was like, hey, we're kind of gonna, you know, Silicon Valley, AI companies, big tech, they can do their thing. But like, we're the play for like the rest of the, of the country. We had this very big vision and honestly stopped doing it because it works so well. And like we kind of, you know, had some exits and then basically I think this is like a, still a very promising model. So they liked the idea that like they could transform entire sectors. Like people talk about that a lot in VC pitches, but like, to really credibly have a shot at doing that, at transforming a sector is very special, very unique. And then regarding insurance, I think we were able to line up this like, arrangement where we had data from a company that had at the time, 20 billion in premium annually flowing through it, millions of, you know, business customers, or I should say, you know, they're like 6 to 7 million SMBs maybe in the country. I might be a little outdated. And they had a meaningful chunk of these businesses. I mean, just really they bought 100 brokerages a year, so they had really big footprint. And I think the combination of industry transformation potential, incredible data and just like a team from the very beginning, we had just like our first few hires were just like stellar, stellar people. Meant that we could drive an insanely high talent bar.

Yeah. How much of your success do you think is because you hire great people.

I think, I think that's, that was the primary factor in this case because I think not only did that mean we were able to grow quickly, but also it meant that it would be very hard for the incumbent to replicate. Yeah, and we, we, I think we had an insight early on that like the quantity were going to be merging with AI and this is obvious now but like you know, the quant and self driving worlds are kind of like the same talent base. At the time, I don't know that it was as obvious. And so we were, you know, we, we were going to places like you know, jump trading, high frequency trading or two Sigma and realizing that hey, they look at time series data. They do, they apply machine learning to it. We insurance kind of looks like that. Different timescale, different granularity, so on and so forth. But it was like, hey, we could put those, that talent together. That's what we did. And you know, just it would be impossible for an incumbent firm to hire those people, in my view.

Yeah, well, you know your story kind of reminds me in a lot of ways you have like a lot of similarities but decades later with the founder of Renaissance because you also were at a big investment firm that utilized AI and he also had a DoD background where he did a bunch of math for it. Not machine learning and AI, but a different type.

Yeah.

And then eventually sort of created his own ecosystem in the investment realm. But you're doing it now in a different space. Well, you're kind of full circle now in investments too.

Although I will say he's significantly smarter. I can't compare myself to Jim Simons but, but he who passed away recently and Renaissance is a legendary firm. But I think, yeah, I mean he took kind of outsiders. Like he didn't hire financial people. Categorically he didn't. He hired like mathematicians, geometricians like himself and they, they just won. They just found a formula for winning. Created a really unique culture and all that. So yeah, I mean I was never, I, at one point I was managing a medium frequency stat firm, statistical arbitrage firm. So I was doing some quant stuff but never like in the kind of like Renaissance tier. I mean nobody is really.

Yeah. Although who's that other big trading firm that's out of New York that's they have a similar model and all the partners I remember like inappropriately because you know me by now, I remember sitting next to them and they were all at AEI so that that conservative think tank and there were like Seven partners, all at aei. Yeah. And I remember sitting next to them and one, they were outsiders like everybody else. There's like suited, buttoned, you know, one had like a ponytail, you know, kind of like different outfits. And, and you have to donate a lot to go to AI. So it's like one firm, seven people, all donating.

I'm like, yeah.

Does the firm sponsor it?

Yeah.

And they're like, no, we all independently. And I was like, you guys make a lot of money, huh? Like, what exactly do you do? And they were saying that they're one of the only other firms that really hires, like only physicists, only mathematicians. So I never thought about that because when I was in finance, that was really not the case. Yeah, I mean, you hired traditional finance investment professionals, even at Goldman, you know. Right, Okay. I want to get to two things, actually. I want to round out here. Like, let's say we're going back and we're talking to young Brendan back in the day, and they're listening to your success and they're like, man, I want to figure out AI. Or maybe I just want to make millions or hundreds of millions of dollars. What, what are some of the resources today you think exist where you're like, if I was a young man, you need to be reading this, listening to this. You always are the one of the people that I go to that I'm like, what, what, what philosophy are you reading now? So like, what would be your list to a young man who wants to be successful in the AI age?

Yeah, I'm gonna have a very unusual answer to this question and not especially practical. I'll explain what I mean by this. But so I think, you know, after selling the companies, I discovered philosophy. I took one philosophy class at mit. I did not like it. And so it came to me, like when I was ready for it. And I realized that that gave me the most interesting, clearest grounded sort of position on AI that I could possibly have. And so what I, what I did is I looked at kind of like technology through this very unique lens of the history of ideas in a way where, you know, Aristotle 2500 years ago wrote about the possibility of self guided machines. You know, the enlightenment thinkers were much more ambitious than even our, more than even our, like really ambitious Silicon Valley folks. I would say, like Francis Bacon talking about the mastery of nature and laying out that project for the world. Or, you know, critics of technology who, I think we should, we should hear their voices. I would argue the best critics of technology died like a long time ago. They're people like Alexis de Tocqueville or C.S. lewis or Martin Heidegger. And so I didn't know any of that. And as a young man, I had this, like, desire to want to help other people. You know, like, I'm very glad that I was. I, I attribute that to my parents, but it was not tutored. I had not really thought about, like, how do you actually help people? Like, what does it mean to live a flourishing life? So I would say, you know, read those greatest minds, like, and read people who are outside of the technological bubble precisely at the moment when the thing is, you know, the tech frontier is moving so fast.

Yeah. You know, it's interesting you say that, because oftentimes I'll be reading something like a great. And then I'll think about it, and then I'll write a tweet about it, and then somebody will say, oh, I have that tweet. That's, that's, that's my tweet. And then I'll go back and quote, like, one of the grades, and I'll be like, no, no, this is actually based on Dante's Inferno. And really, the idea is not. The words aren't exact, but the idea is based on the grades. And we have such a short history now of, of reading real literature. I think that people think these ideas that they have are their own often. And so I, I, I find that point really well taken. It's like, sure, read Mark Andreessen. Sure read Paul Graham. But also go, like, three steps deeper to where they got the original ideas from, too.

Yeah, I think Lord Acton has a quote. This is kind of meta that. One of the most, I guess, frustrating things is knowing the, the provenance of an idea, because it is the case that, like, we have original thought, but in many cases, we've been shaped unthinkingly. You know, we've, we've accepted a kind of frame, for better or worse. And so philosophy tends to make that stuff explicit. It's like, no, no, what you're arguing for is a kind of. It's a Nietzschean idea. You know, it comes from Nietzsche or, you know, this is something that Aristotle articulated in the politics or whatever. And that's helpful not just for kind of, like, intellectual showmanship, but it's really, really helpful then, because what you do is you get a, you have a little hook then into these debates that have been going on for 2500 years. Like, I was arguing on Twitter today with a guy who.

Never mind.

Yeah, but I do it Sometimes. And I like it. I admire people like Emmett Shear who like just, you know, authentically argue with people. Not like, you know, knife fighting kind of things.

Unemotionally.

Yeah. Quasi rationally earnest. Yeah. And. But anyway, this person was talking about a proposal in which we all get to live on our own planet, essentially, like in our technological future. Right. And he was sort of trying to obviate or abandon the problems of political philosophy. And without, you know, having read through the debates that we've had for centuries about, you know, the political problems that this entailed, I. I would have had a lot less ground. I would have just maybe intuitively been like, this is not a smart idea. But I wouldn't have been able to like, really understand, like what debates it even, it even raises.

That makes sense. So if you had to give somebody a really tactical, Internet built list of things to read as a young man, what would those be? Like? What would be the seven books that you'd be like? Start here and then expand.

So you have to read Plato's Republic and you have to do it, I'm convinced, with a tutor or in a group. I say this because I read it three times alone in a corner in a chair after putting my kids down. And I was just like puzzling over it and I didn't get quite as much out of it. Right. And I'm, I'm a sharp guy, but like, it's very hard to read these texts alone. So then I started going evening classes at St. John's College. I started doing tutoring with a guy named Mike Millerman who people should check out. He's great. And that unlocked it for me. And why I say you got to read that is that you know, someone, I think maybe Bertrand Russell, I forget, but said all of philosophy is a footnote to Plato. And it's kind of right. Like these debates, he frames them really genuinely, really broadly. So I would say read that. The next thing that I would. That mattered a lot to me to read was a book called the Constitution of Liberty by Friedrich Hayek. And this is, I think, a good argument for why liberty is important. Like, you know, why freedom is something we cherish. He lays out an argument based on the usefulness of liberty. Particularly sacred is this essay called the Creative Powers of a Free civilization. It's chapter two and you can read that. It's like 16 pages or something. And that's wonderful. I would recommend Democracy in America, but it's very long. I would. There is a Harvey Mansfield version, just to get very tactical. That's a short version. It's like 120 pages. I would read that. It's faithful to it. Like, definitely read the Tocqueville, but start with the Mansfield version and go, go for that. Tocqueville does not write about technology. He spends 10 pages on it in a very esoteric way. Calls it the poetry of self reflection for democratic man. Weird. Awesome. But he writes about trends that merge with tech and they're so, so powerful. So you gotta. You gotta read that. Those. Those are the books that I would probably start with. And again, on the first one, gotta get a group, gotta get a tutor. It's just. It's necessary, I think, to get what you need out of. Out of Plato.

Yeah. That's so good. Especially in the age of, like, lack of attention and inability to.

Oh, there's one more. Oh, please. Sorry. The Use of Knowledge in Society. This is an essay again by Friedrich Hayek, who I think is wildly underappreciated for his tech insights. He is an incredible thinker on the use of knowledge in society, which obviously AI affects. So I would read that essay. That's an incredible one to read right now.

Yeah, it's so good. So if you're. If you're hungry and desiring of more of this information, I highly recommend you go to Cosmos. I also think I'm excited for the YouTube channel. I also think they should follow you on X where you've been out there more. That's where I first saw your idea about Entrepreneur Nation versus Employee Nation.

Yeah.

And I know you're gonna put more stuff out there. So what's. What's the X where people should follow?

So it's M. Brendan 1.

Yeah. And so we're gonna work on your naming convention.

I'll warn you. Yeah. My. My stuff is all original. It's all very. It's hot takes and, you know, but it's like, it can be a little bit low tam, so we're working on. On that. But it's. It's cool. I mean, I love it. It's. When I started engaging authentically on Twitter, it like, it actually, like, improved my. It was a key part of my intellectual journey, actually. So. So I really love it.

Yeah. It makes you synthesize really complex things into really simple things. And you're rewarded by. If it resonates with somebody directly by the feedback loop.

Yeah.

Which I actually think is really beneficial. And. And often I think people try to go super highbrow with the things that you talk about, like AI and you do A really good job of saying, well, yes, I. You there. We can go to that level. But actually, bottoms up. We need more people to utilize this, not just the few on top who are already there.

Yeah.

So I hope you do more of that. Brendan, it was such a pleasure to have you here.

Likewise.

I just psyched you out that we were going to be done. But I want to do my favorite part, especially in this age of, like, humanness. So at the end of the podcast, what we do is we give everybody a little postcard. And on the postcard, we give you a few minutes to write to young Brendan. So you pick the age, but something before you feel like you came into being a man, a little message, like, what would you tell him today? In the world that we stand in today, it can be in the context of anything we talked about or anything else. And then you're gonna read it, and it's going to be this. Like, the idea is every time we do this, it's a reminder that, like, even people who are in positions of power who are on podcasts like this, like, they have that young voice inside of them, too, that maybe didn't know.

Okay. All right. So right around the time when I shifted from kind of pure play tech to philosophy, my wife shifted into mental health counseling. So this is her passion. What she caused me to realize is that a key factor in my success was guilt. It was this idea that smart is as smart does, my mom always say. But it drove me every Saturday morning, every Sunday morning to, like, if I got up, I was like, I gotta be working. I gotta be working. And what I realized is that the parts of you that are ingrained, that made you successful become really, really strong. And then there's a little bit of an atrophy of the other side. And so leisure reading, that was never developed. I couldn't do it. I would panic. To this day, I haven't meditated, actually, still, ever. So I had to develop that. And I told myself that this. Smart is as smart does. Part of you is so dominant. Cultivate, develop like a muscle, the crucial complement of leisure. And then I drew a cross section of my brain showing.

Let me see it. What's it look like?

It's got the guilt part here and the leisure part here, but I should show it growing because it's a redemptive story.

Is it?

Yeah.

Are you getting better at leisure now?

Yeah. I'm not perfect, you know, but I think I. I've. I've come to realize that it's. It's A crucial part of, you know, a full human life.

I love that. Brendan, thank you.

Thank you.

I hope you were listening as closely as I was to this episode because I think people will become billionair billionaires this year, next year, the next five years as one person companies from AI. I think there will be billion dollar one person companies created from AI from what I have seen firsthand. But maybe more than that. I think we will be able to have employees for the first time ever, even when we don't have a business in the form of AI. I think all of my employees will have employees going forward. I think we could run our own mini businesses inside of our company. I actually, after listening to him, some of the things that I took away the most were that he's pretty optimistic about AI. As long as we keep doing the thing that we are doing together here, which is questioning everything. I call this company Contrarian Thinking, because I want us to think. I want us to hold up our brains and actually analyze the world around us. And if we don't do that, I think AI takes over entirely. And AI is really just letting other humans who build it hijack your brain, tell you what to think, tell you what to do and take over your life without ever having a war. The war will have already been fought and it's with a keyboard. And we welcome the enemy into our home. And so I don't think that AI is the enemy, but I do think if you are not really thoughtful on being the type of human that pushes back on the things that are told to you at face value, then you need to start becoming that. And the way that you do it is by listening exactly like you are here today, and then by questioning everything we say, just like you question everybody else. And I know that you question us sometimes because only 53% of you are subscribed. So you loved this podcast. You got all the way to the end and then you're not subscribed. What do you want? You want the AI robots to become your overlords? So if you guys want us to continue bringing you the best content we humanly can, to obsess over every guest you want and to continue to ask the tough questions, please subscribe to the channel. We want to get to know you and build a real decades long relationship with another human. By the way, no robots here. And I want you to know that you are a big deal to me and it is an honor for me to have you on this podcast with us today. So please make sure to subscribe and I'LL see you next time. Hey guys. If you've ever thought about buying a business, we've built what I think is the best acquisitions and business buying community and education curriculum in the world. If you've ever thought about wanting to buy or own a business, or if you want to add more businesses to the mix, it's called the Contrarian Community. And what this is is the goal is we give you the three things that the best business buyers use, your own advisory team, your own investment committee, and a deal team. We get together each week to review deals, live and beat up all the deals that you're currently looking at while you simultaneously learn the best way possible, which is called modeling, by seeing other people put together deals. This is how private equity buys businesses. This is how investment teams work and we're stealing the methods from Wall street and giving them to you. If this is interesting to you, go to click the link and you can actually talk to my team direct about if this is a fit or not. We can help guide you. The link is in the show description.