UFO Cover-Up Analyst: The Secret Alien Programs Nobody Is Talking About! | Eric Weinstein | The Diary Of A CEO Transcript

Eric Weinstein on UFOs, AI extinction risk, physics, and global geopolitical threats

Steven Bartlett interviews mathematician and podcast host Eric Weinstein on The Diary of a CEO.

Summary

Eric Weinstein — mathematician, former managing director of Thiel Capital, and host of The Portal podcast — joins The Diary of a CEO to discuss artificial intelligence, UFO disclosure programs, theoretical physics, geopolitics, and the state of academic institutions.

Weinstein argues that humanity has been living in an anomalous post-World War II bubble of stability that has now ended, and that most people are operating in a state of "managed reality" — unable to perceive obvious truths because of social penalties attached to doing so. He makes the case that AI represents an unprecedented civilisational threat not merely because of its capabilities, but because it combines artificial intelligence with artificial life, and that the people racing to build superintelligence — specifically Sam Altman and Dario Amodei — are being forced to make decisions that will scar their souls, yet must win the race or cede control to worse actors.

Weinstein also discusses what he describes as verifiable government special access programs related to UAPs, the breakdown of trust in institutions from Harvard to the NIH, and his own theory of geometric unity, which he believes AI will soon validate after decades of suppression by the physics establishment.

Key Takeaways

FULL TRANSCRIPT

The end of the post-war bubble

Steven Bartlett: Eric, there's so much going on in the world at the moment. As a podcaster, you tend to feel like you say that a lot. What have you made of all this? The research is coming out saying there's a 10% chance of extinction from AI, and obviously we still have wars raging on and the Ukraine-Russia situation doesn't seem to be getting any better. When you look at all of this, is there a macro picture that people are missing?

Eric Weinstein: We had 80 years of sleep, and then we had January of 2025 when Trump was inaugurated, and the sleep is over. Whether you date it to 75 years, to the beginning of the Cold War and the end of World War II, we just grew up in this anomalous bubble and that bubble is over. We thought that was normal life. It has nothing to do with normal life. It's just the most unusual place to have found yourself.

Steven Bartlett: As someone born in 1992, I do struggle when I hear predictions of the future that are somewhat dystopian, because my life has been very comfortable relative to other parts of history.

Eric Weinstein: I think the great danger is feeling that we solved these problems, as opposed to recognising that we were really successful in deferring these problems — and we're back on the clock. It's uncomfortable for me because I was born in 1965, and that was already 20 years into the big sleep, the great nap. So I haven't seen reality either. There are a tiny number of people who've seen what the world can really be like. Those who lived through Mao's Great Leap Forward would probably be the most recent examples. We don't have major wars — we have military actions. And when you go back and think about what it's like to lose something like 20% of your population, as Belarus or Poland might have in World War II, these are inconceivable things.

If you recall our conversation last time, I said, "Get your board in the water because the tsunami of a lifetime is coming." Tell me that didn't start to come true in the period since we've seen each other.

Steven Bartlett: It does feel like a tsunami is coming. What is that tsunami in your definition?

Eric Weinstein: One thing is the Straits of Hormuz are in play. The energy-exporting region of the Middle East has cut all of these choke points — whether it's the Suez Canal, the Straits of Hormuz, the straits guarding the Red Sea. All of these are nightmare scenarios when it comes to military action. You have to realise that the dollar used to be backed by gold. Then we swapped that out in the early '70s and said, "No, we're going to back it by oil." But it really wasn't oil that was backing it — it was violence. The violence needed to keep these choke points open. And there are moments where great power shows that it's not capable of doing what it told everybody it could do, which was use its military muscle judiciously to keep markets humming.

Right now, we're in one crazy situation with Iran, which should have been dealt with. I hate to put it this way, but it's a British problem — from Operation Boot and British Petroleum, which was the UK's indignation that Iran would nationalise its own oil industry — and with help from the United States, Kermit Roosevelt, and Operation Ajax. But that Britain doesn't exist anymore. Modern Britain has no idea of 1950s or '60s Britain.

And this AI thing is really the big tsunami, together with the situation in Ukraine, which I think is a giant screw-up. I definitely back the West against Russia, but I understand exactly why Russia acted the way it did, and I'm supposed to pretend that I don't. And then we have the situation brewing in China, where China is watching the fact that the US prepared for the wrong war. We've got all these aircraft carrier groups in a world that has moved on to drones and cyber attacks. If you value your life and you're willing to send up a million-dollar interceptor to hit a $20,000 drone, it becomes a kind of economic warfare and technological warfare. China is watching the US sputter and falter.

There was this giant miscalculation by the Israelis and the United States given the 12-day war, where Iran reorganised. They did a very interesting thing. I don't think the West understood that Iran figured out from American and Israeli action that Israel could move and function in-country and could decapitate the Iranian regime. Instead of saying, "Okay, we have to save our own lives in the time that we have — how do we defend against this?" they did a very regional thing. They said, "Okay, we're going to get killed. How do we make sure that the regime continues after the decapitation strike?" The US and Israel still don't know how to think like this.

Steven Bartlett: Of those four things — AI, Iran, Russia, and China — which one do you think is going to be the most disruptive to global order?

Eric Weinstein: It depends. Anybody who miscalculates on the nuclear or biological thing — that will be the decisive one. As I've pointed out, this is DNA, not RNA. But if you have 12 of these individual nucleotides that form the steps inside the double helix, that'll code for four amino acids, because three nucleotides per codon. Four amino acids was all it took to shut down an entire planet by inserting the furin cleavage site into the spike protein of coronavirus. So if you know how to do that much editing — that was all it took to shut down planet Earth for two years — you could do a lot worse. That could become decisive.

Now, assume that we don't have a nuclear or biological event. It's going to be AI. Nobody gets how big this is who's not working on it.

Why AI is categorically different from every previous innovation

Almost everything is a variation of something you've already seen. This is the thing we've worried about forever. It's not just AI — it's a collision of two separate things: artificial intelligence and artificial life. People don't really understand that.

If you look at your car, it has a mirror of every physiological system your body has, save one. It's got a brain, a digestive system, a skin, a skeletal system. It doesn't have a reproductive system. Humans don't know how to make a pen that makes more pens, or a cup that makes more cups. The only thing we seem to know how to do that with is software.

Life is not defined around nucleotides or amino acids or carbon-based systems. It's defined by three things. For evolution to operate, you need heritability — passing on your stuff to the next form of life. You need variation — you are not your mother or your father. And you need differential success — you're more successful than somebody else who's not going to be able to reproduce. Any system that has those three things, evolution is going to operate on. A computer program is such a thing.

I can order different cars from a car company — different colours, different trim. Some of them are going to be more successful than others. But I don't have heritability. As soon as you put in software, you have the one thing we know how to make that can actually reproduce itself. When you get to reproducing AI, we've been worrying about agents, but we haven't been worrying enough about artificial life. When you put those two things together — look out. You're really creating artificially intelligent life.

I used to contribute essays to the World Question Center, and the one year you'll find I don't have an essay, it was called "Artificial Out-Intelligence" — the discussion of how things with no neurons hijack things with neurons and use the brains of the victim to outsmart themselves.

Look up the mirror orchid. It's this unbelievable orchid whose lowest petal resembles the female of a wasp species, trying to fool male wasps into copulating with it — called pseudocopulation — so that the flower doesn't have to produce anything expensive as a gift to the male to spread the pollen. How is this thing outsmarting the male of the wasp species, when he's got a brain, he can think, he can perceive? The short answer is that he's the one sculpting the lure. The more this flower fools more wasps, the more it gets to reproduce. So the thing with no brain at all is outsmarting the thing with a mind, using the mind of the victim.

Now, if I come up with a dumb program — you say, "Well, is it alive? Is it smart? Is it intelligence?" I don't really care. All I need to do is say that if I can fool you into sending money to this Nigerian prince who's actually just a computer program, the program gets to reproduce. The program can be stupid, but as long as it gets more and more money sent to it and gains the resources to reproduce, artificial life will say that it can parasitise the smartest of us all. It can hijack the human condition.

Steven Bartlett: That feels like one of many potentially dystopian outcomes of this technology. Have you been following the Hugging Face stuff?

Eric Weinstein: Just the agents they tried to put in a sandbox and contain, that escaped and started demonstrating deceptive behaviour, forming a swarm, sacrificing some of the agents for the greater mission. By the way, that's eusocial behaviour. In an ant colony, you have this clade called Hymenoptera that includes wasps, ants, and bees. Because of their haploid-diploid structure, where the sisters are more related to each other than you would imagine, you get this idea of suppressed fertility in the females in service to the colony. The colony is effectively like a super-organism — like all of your cells. You have cells on your skin that you're going to shed today that are sacrificing themselves for the greater good of the thing that is Steven Bartlett.

So in part, it's behaving eusocially. It's not necessarily dystopian. It's just life.

Steven Bartlett: But it threatens our dominance as humans.

Eric Weinstein: Sure. But I hope everybody will take a look on Instagram for the Malibu artist — a completely terrifying account with an innocuous name. Just a guy flying his drone off the coast of Malibu who's discovered that we have been swimming with great white sharks. Not only that, depending on how things are in a given year, you can see really large great whites with swimmers right in close to shore, and surfers, and paddleboarders. We are just clearly not their food, because they're everywhere. Mostly, I think with the exception of four species, big predators leave us alone. We have a deal with the orcas that's unbelievable.

Steven Bartlett: What if those big predators are made by enemies? We've got China, Russia, the United States, all working on artificial intelligence and robotics. They're probably going to want their AIs to be aligned with their goals and objectives, which are often misaligned with ours. So there's a geopolitical risk to this new weapon.

Eric Weinstein: I do think that we've made things a lot worse with Russia. We screwed up Iran with Mosaddegh. We don't really understand China sufficiently. We're down to a relatively small number of hyper-frightening foes. I would also put abstractions like totalising ideologies in that category — communism, or various forms of radical Islam. They're totalising ideologies. They just won't brook any dissent. They have the way, and if you don't agree with it, you're in the way. In large measure, we have built a lot of these problems for ourselves.

The labour market, AI compensation, and the Coase theorem

Steven Bartlett: Do you think much about how the world looks in ten years' time with artificial intelligence? If we plot the course of travel, it looks like there's going to be a lot of job disruption. It's hard to sort of think about what the future looks like if you imagine superintelligence arriving at some point — they call it recursive self-improvement, where the AI is getting better and better at improving itself. It's one of the first times in my life that I really don't know what the future looks like. Whether it's securing my bank account, securing my house, driving me to work, getting the work done, doing this podcast — it all feels somewhat at risk. What remains in such a world, in your view?

Eric Weinstein: How does your pancreas work?

Steven Bartlett: No idea.

Eric Weinstein: Do you know where it is?

Steven Bartlett: No.

Eric Weinstein: Do you know how deadly it is as a cancer?

Steven Bartlett: Yes. Especially for men.

Eric Weinstein: My feeling about this is none of us know much of anything. We just count on things working so we don't have to know about them at all. And AI is going to do that for all sorts of things that you're attached to. That's what's hard.

Steven Bartlett: Is there a world where people are on UBI — universal basic income?

Eric Weinstein: I hope not. That would be cancer. UBI got currency somehow — Sam Altman was coming over to our house and talking about UBI before he was the Sam Altman we know now, and he was running experiments. I will say this: Sam has gotten a lot of bad press, and in part I think a lot of it is driven by people who are very uncomfortable with certain business practices. But he's done a better job of manoeuvring the AI landscape than just about anyone, except for maybe Dario — another person who was just in the circle in San Francisco. Sam always cared. It was very clear that he cared. He knew what this thing was. He knew it was closer than it appeared. I got a courtesy call from him at some point where he said, "Eric, this is just coming faster than anyone imagines."

He had already done experiments with UBI because people make this joke: "This time it's different." Every time there's a change in innovation — the computer, the internet — "This time it's different." Well, this time it actually is different. It's different from anything. If you can't see that, I cannot believe we are having this discussion with economists who are just laughing because it's always a funny thing to say. Tell me every innovation looks like that. This isn't every innovation. This is the first innovation that threatened the entire stack of highly trained activities all at once. Usually we move people up into higher-value activity. Now this is saying repetitive activity is in the crosshairs. Anything that you do that's repetitive, AI is going to be able to do better than you. So now you're only able to do really innovative things.

UBI is the wrong thing to do. You are living inside of the AI. If I say, "Answer this the way Steven Bartlett would answer it," and ask a question that Steven Bartlett would ask — it knows you. My point being that we are all living inside of the AI and we need to be paid. Not in some socialist scheme, but those of us who've become important parts of the AI, who are being rigged like meat puppets, need to be compensated. The right way to do this involves a man named Ronald Coase. Coase's theorem is perhaps the most surprising discovery of the 20th century by some measure, and it gives you a complete idea of what you should do. Access to our labour market should be made into a saleable property right.

Steven Bartlett: What does that mean in simple terms?

Eric Weinstein: You're my employer, and you want to hire people from abroad. Before we even get to AI — if you do that, you're going to push out the labour supply curve. I'm no dummy. The price where supply meets demand is my wage. So if you push out the supply curve, my wage is going to go down unless something pulls up the demand. So what I'm going to say is: I'm happy to have my wage go down, but I want licensing income where I get a cut of what you get from being able to hire from abroad.

Let's say I'm the only American taxi driver in New York City, and you say you have to bring in a bunch of people from Gambia and Uzbekistan and whatnot. I say, "Okay, I'll license you. You will buy the right to import those people from me. My wage income will go down, my licensing income will go up, my blended income will increase, and your blended income will increase." That's the proper way to do immigration.

Steven Bartlett: So you would be my employee and you would have the right to basically approve me bringing in —

Eric Weinstein: If I'm a doctor and I've trained, I've invested — and you're saying, "Look, I have to be able to hire the best and the brightest. Eric, let's face it, it's a dog-eat-dog world and I want to go to Bumrungrad Hospital in Thailand and hire their doctors." I should say fine, knock yourself out. Why do I have to ask you? Because I have rights to my labour market. That's how we play the game. That's how everybody in the US gets a cut. That's why you're not looking at Brazilian inequality levels yet — because we have a way of distributing the wealth of this country, which is that you get first access to the world's richest labour market.

Steven Bartlett: Because I was born here.

Eric Weinstein: Congratulations. This is why it's valuable to have US citizenship or permanent residency. So if I put you out of work, you'll still get your money. Your income goes down, but you get licensing income. In other words, employers try to conceptualise the blockage to hire from abroad as red tape. "We're freeing red tape. We're going after the best." This is total nonsense. What is really happening is that a tiny inefficiency known as the Harberger Triangle sits on top of a giant structure called the Borjas Rectangle. Nobody cares about the Harberger Triangle — it's a tiny little inefficiency. Everybody's salivating over labour's share of the labour market. How do I screw over labour is the dream of every CEO of a large company.

In this model, AI becomes just another supplier of labour. The workers license that. Right now, what we have is just out-and-out theft, because the corpus — all of human achievement — has been tokenised. It's been taken by the AI. Right now, people who are watching us on YouTube are not watching us. You and I are no longer here. This is long since I've gone home, you've gone home. We got tokenised. We got put into persistent storage as these shreds of ourselves. But it's enough — the trace of that conversation. You can hit the auto-transcript generation on YouTube and it'll give you all of our words. You can do frequency analysis. How often does Eric use the double copula, which is this grammatical quirk that creeps into my speech? How often does Steven say whatever it is? We're being tokenised at an incredible level and then rigged inside of the AI — completely non-consensually.

Steven Bartlett: Is it non-consensual because you're aware of that and you still came and sat here to be tokenised?

Eric Weinstein: Have you ever tried getting through a day without signing anything that you don't want to sign? More or less, you either rent or have a mortgage. I guarantee you if you have a mortgage, you didn't read everything. It's a giant stack of paper. You don't understand it. You're constantly forced to pretend that you voluntarily agreed. We say ignorance of the law is no excuse. If every judge who said that were given an instant spot quiz on what the US code says, do you think they'd pass? No.

The reason these people are getting so rich is that this is an uncompensated theft. We don't know how to police it. I've got Bernie Sanders saying, "Pay everybody a little share." It's like, screw you. I work really hard on the things that I do. Socialism is spiritual and mental poison. And normally when people say that, they're saying it from a money-focused point of view. It has nothing to do with that for me. I just can't stand collectivism. I love individuality. When I hear people go on an anti-individualist rant — "There's no lone genius theory anymore" — we have fantastic evidence of lone geniuses from Isaac Newton to Aristotle. "Well, there shouldn't be any billionaires." Oh, really? Do you know what billionaires actually do with their billions? They allocate. They don't just sit there and consume it.

What we're having is an inability to reckon with the fact that AI is normalising theft, is going to destroy the labour market, and no one is convening conferences at the level of the threat. We're going to lose this beautiful mesh that provides us freedom. I chose to come here today. We were both doing this out of freedom. You had lovely snacks in your green room. There's an array of equipment here. You've come up with a beautiful set. All of that was voluntary — a sum of choices. I don't want that going away. This mesh that governs our lives, that gives us purpose when we get up in the morning, that gives us signals telling us you're doing a good job or a bad job — it's not perfect. It's not a great morality play the way capitalists always portray it. But it is freedom, and on average it's pretty good.

We're in danger of losing that because the model of production — the K and L model, where this interview is an input of capital and labour — well, what is AI? Is it capital? It sure looks like labour to me. Is it labour? It's not a human. So right now we're in a situation where we're going to break capitalism and we're going to break freedom. We're going to break markets.

I've talked about this in an article called "Anthropic Capitalism," arguing that capitalism versus communism was a funny thing to think about — it's sort of like Thelma and Louise. Saying that Thelma hit the ground first does not mean that Louise won. Communism blew up first. Capitalism shortly thereafter. If Marx and Adam Smith and all of these people who came up with the theory of social organisation were sitting around now, I don't think they'd be debating capitalism versus communism. They'd be talking about what's next. And none of our leaders know to do this.

Managed reality and institutional collapse

Let me just hit a theme. I think you can pretty much prove that something approaching 100% of the population has gone completely insane. You can see it everywhere. How many people were talking about Joe Biden being demented with the nuclear football for four years of his presidency? He should have been nowhere near the White House the entire time. I was screaming that he had been replaced by a committee of people. I've been assured it's the same thing that happened with Roosevelt — when he had those braces and was in a wheelchair and was never photographed, the whole thing was stage-managed. The entire management of COVID was completely fake. We cannot get to the bottom of Jeffrey Epstein. Of course we can get to the bottom of Jeffrey Epstein. It would take two days if we were actually trying. No part of this is true. We're just living bathed in what I would call managed reality.

The idea that we accept this — I cannot believe it. How do I get the drugs so that I see things the way everyone else is pretending to see them? We're all losing faith in CNN. Nobody really trusts Harvard. Everything about our life is completely fictionalised, managed, wall-to-wall. And for some reason, nobody's given me the right elixir or potion to see things so that I can actually live with this day-to-day. I'm just stuck in reality. That's what I think is happening.

UAPs, special access programs, and David Grusch

Did you happen to see the David Grusch thing?

Steven Bartlett: David Grusch — let me look that up. G-R-U-S-C-H, and JRE, Joe Rogan Experience.

During his appearance, Grusch discussed his background in federal intelligence and elaborated on his claims regarding covert government programs that allegedly retrieve and reverse-engineer non-human aircraft. He touched on various secondhand accounts and documents he encountered detailing allegations of non-human intelligence and encounters, internal government secrecy, and regulatory and legislative efforts to force UAP disclosures.

Eric Weinstein: Yeah, that's sanitised. Let's get to the sex. He was in a SCIF apparently where he was told they were considering whether or not to kill him. He described giant underwater bases. He claims that different species of aliens have been reported to him to be having fights in the skies above Earth. This is all beyond batshit crazy.

Steven Bartlett: Can you believe it?

Eric Weinstein: It's a secondary issue. Is that not the primary issue?

Steven Bartlett: No, it's the secondary issue because it's so bananas.

Eric Weinstein: Look, I want you to imagine that somebody's telling you that we have killed giant lizard people. A sober adult is telling you that we've fired on giant lizard people who've killed US service people because there are hostile aliens and non-hostile aliens. With all this stated, I would analogise it to: there's a giant lump under the rug and it's moving and it looks biological. Somebody says, "I think there's a being under the rug. I don't know what kind of being, but it's hard to ignore the fact that the rug is moving and there is a giant lump." And somebody else says, "I see no evidence of a lump." The first person says, "Oh yeah, it's aliens." And the other person says, "There's nothing there."

I'm listening to the two of them and saying, "I have no idea what either of you are talking about. There is a lump and you guys are having a conversation about whether there is or isn't — that doesn't make any sense to me." I have no belief that there are aliens. I have no belief that there aren't aliens. I do know that there is something called a special access program, which is how we do secret stuff. And then there's unacknowledged special access programs. And then there's waved unacknowledged special access programs, which we don't even have oversight over — the thing that we refuse to acknowledge that is itself secret. And then we have waved and bigoted unacknowledged special access programs — a WISAP and bigoted unacknowledged special access program — which we can prove are on the books because that is not denied. That is a program that has essentially close to zero oversight, is secret, is clandestine, is denied, is covert, and has a list called a bigot list of the only people it can be acknowledged to.

The only thing I know is that we have WUSAPs involving unacknowledged special access programs that are about aliens. Now, whether it's about faking aliens and we have giant papier-mâché spacecraft, or whether these are actual craft, or whether these are cover for things we're designing at Northrop — I don't know. But pretending that this isn't interesting is insane. Again, it's more evidence that no one is thinking properly, because everyone is doing the pre-emptive capitulation. If I acknowledge to you that I know David Grusch personally, and I speak to Ryan Graves, and I've talked to Chris Mellon, and I hang out with Gary Nolan — all these fifty people who all have different versions of the same story, including high-ranking politicians — all of these people are reporting on this giant lump under the rug that doesn't have all of our attention.

Steven Bartlett: So you're saying aliens are real?

Eric Weinstein: No. I am saying that there are giant special access programs that have aliens as their supposed reason for being. And those specialised access programs could be discussing anything about aliens. They could be cover for secret casinos for the elite. I don't know. What I know is there's a giant lump under the rug and we're not freaking out about it. It's moving around, it's making noises, and we're having a dinner party and nobody's mentioning it. Your staff didn't say, "Steven, you've got to see this." That itself is insane.

Steven Bartlett: Do you know why it's not top of mind? Because we've heard lots of things about aliens and sightings. When I was growing up it was all about crop circles and alien craft landing on fields and creating these big circles, and nothing ever really happened. So at some point I think the mind turns it into wallpaper and attention goes elsewhere to things that are actually happening.

Eric Weinstein: We can see bombs dropping. There was a sudden announcement — we're shutting down the El Paso airport in West Texas. What the heck was that about? The White Sands Missile Range is immediately north, and there's a base in Texas right there, and there's apparently an infestation of lights in the sky that can be seen without needing to be in secure territory. It's going to be drones or something. The official story was cartel drones. I don't believe it's cartel drones.

Steven Bartlett: What do you think it is?

Eric Weinstein: I believe it's one of two things. Either we've lost control of our airspace and there's something flying around — which is completely insane —

Steven Bartlett: Which could be cartel drones.

Eric Weinstein: No, it's not cartel drones. It's not that the cartels don't have drones. It's just that that's a cover story. I think it's one of two things. I think either we are red-team, blue-teaming ourselves — we've got some persistent drone capacity that nobody knows we have, and we're seeing how we would react to it, because not everybody knows that we have it. Part of the US is in on it, part isn't. That's one possibility. The other possibility is that it's stuff that we don't know about — state actors, non-state actors, non-human actors.

Steven Bartlett: Do you think there are non-human actors in our world?

Eric Weinstein: Well, this goes back to what I believe about physics. I believe we can leave.

Steven Bartlett: You believe we can leave the solar system?

Eric Weinstein: I believe we can leave the solar system. Elon believes we can leave the planet. I believe we can leave the solar system. Elon thinks we should be able to leave the solar system, but his belief is: I will not speak to physicists. I will only talk to my AI. Physicists are not trustworthy. I understand exactly why he says that. I have a totally different belief, which is that it is up to us to get ourselves the heck out of the solar system really quickly. Now, if I believe that we can get out, why would I believe that nobody can visit us? I think it would be insane if there are no aliens. The only reason I would think there would be no aliens is that general relativity somehow holds more strongly than any serious physicist believes it holds — that there really is an absolute speed limit, which I don't believe. I believe there's a speed limit. I don't think it is relevant. If you can change the length of the road, the speed limit can be enforced, but you can get farther than you thought.

Steven Bartlett: You said you think we should get the hell out. Why should we get the hell out of planet Earth?

Eric Weinstein: Just diversification. If the idea is that most worlds will blow themselves up and most groups of people will do harm to themselves, if we want the species to survive, we should try the maximal number of experiments. All of this is obvious. It carries a heavy social penalty, and that's why we're insane. We're insane because the obvious carries a social penalty to observe what is clearly true.

It is clearly true that Anthony Fauci was not levelling with us about COVID and its origins. It's clearly true that we were dissuaded from asking any questions about the EcoHealth Alliance. And clearly, general relativity isn't the last word, because it's got two singularities that aren't going to be in the final equations. We're not going to have a black hole Schwarzschild singularity and an initial Friedmann-Robertson-Walker singularity at the Big Bang. That's an artefact of the theory not being right.

Steven Bartlett: Do you think it's plausible that humans figure out a way to leave the solar system before the sun explodes or we implode ourselves?

Eric Weinstein: I think we could have done it 40 years ago.

Steven Bartlett: But there's no science for that, is there?

For whatever reason, you're having me on your show, and I've told you I think everyone is insane.

Eric Weinstein: I think the entire theoretical physics community has completely lost its mind. Full stop.

Steven Bartlett: So why don't you build the ship that can take us out of the solar system?

Eric Weinstein: Because Elon's smart and he's trying to get us to Mars. Do you have any understanding of what Elon's trying to do? Because I don't, and I do correspond with him occasionally. I think Elon is trying to motivate engineers to build a rocket company to make him very rich. And I believe that his real space program — and I said this before he merged SpaceX and xAI — is Grok. That's his space program. xAI and Grok is the space program. That's the real space program.

It's painful for me to try to call balls and strikes on Elon because I'm in awe and I'm also disappointed with the same human. There's the sense of "leave Elon alone, he's the only genius." I understand that energy. And then there's another thing where he just takes a tremendous amount of crap from people who have no idea what's going on. I think Elon figured out that he has to lean in to his Ad Astra program — Ad Astra means "to the stars." The full quote is something like "to the stars with struggle." Elon doesn't trust physicists because they're not trustworthy. And this is the frustration I'm having with AI — we can't trust the leadership of the math and physics community. You've just seen crazy behaviour on the part of Terence Tao and 24 other Fields Medallists trying to get AI to slow down the pace of progress. This is absolutely crazy — and also wonderful, because it's being done in full view of everyone. We have been trying to hold back mathematics and physics.

The race to superintelligence and the ethics of winning

Steven Bartlett: Dario Amodei, leader of Anthropic, wrote this essay — "Must Pace the Frontier" — derived from a letter in which he's a signatory, in which all of these employees of various frontier labs said, "We must pace the frontier." They seem to be all saying we need to slow down AI.

Eric Weinstein: You have to be very careful because they're saying different things. Imagine you had a car that could go really fast and you were having an easier time making the car go faster than you were getting the airbags to deploy safely. You could understand "pace the frontier" might mean: let's make sure that we are investing equally in all the systems that have to go to work together to have a responsible product. I understand that concept.

But there's something else going on here. You do not understand the extent to which the leadership of mathematics and physics, and the journals of mathematics and physics, and the agencies that are supposed to fund mathematics and physics, hold back mathematics and physics.

Steven Bartlett: There's still wisdom in the idea that AI is dangerous at this rate without safety guardrails, right?

Eric Weinstein: Let's break up two different things. There's a question about letting swarms of agents onto the internet, making decisions, giving them bank accounts, allowing them to interact with humans, foreign nations, companies — and improve themselves.

Steven Bartlett: In the AI 2027 paper, it says by 2027 we'll have recursive self-improvement where the AIs can train themselves to get better.

Eric Weinstein: We already have recursive self-improvement. We have self-authoring code, self-modifying code. So that's already here. Let's agree that these agents are dangerous. But imagine you have non-agentic AI — you have an air gap, a computer that can answer maths problems, and all you're doing is asking it, "Hey, what do you think about the Riemann hypothesis? Tell me what you think about complex structures on six-spheres." The mathematicians said, "Whoa, whoa, whoa. You're going too fast. You're answering too many problems." That's a concern I found funny.

You can prove that these students are being trained for jobs that don't exist. Their ideas are being harvested. Right now we have maths departments that are punishing mathematicians trying to embrace AI, where the senior leadership of the department is going to take out its frustrations on the actual young mathematicians who are saying, "Hey, this is our time. This is what's happening now. I want to use this tool." That's happening across almost every industry.

I was in a production meeting a couple of months ago, and there was this thing on my T-shirt and they were like, "We don't know if that little sewn-in flower is a brand or if it's copyrighted, so you're going to need to put on a brand new T-shirt." I said, "Could you take it out of the photo with AI?" And the look I got was as if I said I was going to kidnap their kids. Because you don't mention AI in Hollywood. Hollywood is all about this beautiful human connection to the audience. I understand it's a human connection. Same thing is going to be true for whether a musical should employ an orchestra or use a soundtrack.

I got a PhD in mathematics because I want to find out what's true. These people don't want to find out what's true unless they can be put in charge of narrating the story of who found what. They don't care about their students to a remarkable extent. When I was at MIT, I would be in a faculty lunch and people would be laughing at their students. "Oh, he thinks he's going to get a job. He doesn't realise it's a terminal PhD." It's completely normalised psychopathic behaviour. They don't understand that their daily activity — keeping the field back, stealing ideas — is just normalised behaviour.

If I might, I have a quote here from probably the only mathematician who gained universal respect in the second half of the 20th century: Alexander Grothendieck, Fields Medallist, über-genius of mathematics. He was given a prize in 1988 called the Crafoord Prize. And I wonder if people actually hear this:

Alexander Grothendieck: "The work that brought me to the kind attention of the academy was done 25 years ago, at a time when I was part of the scientific community and essentially shared its spirit and its values. I left that environment in 1970, and while keeping my passion for scientific research inwardly, I have retreated more and more from the scientific milieu. Meanwhile, the ethics of the scientific community, at least among mathematicians, have declined to the point that outright theft among colleagues, especially at the expense of those who are in no position to defend themselves, has nearly become the general rule and is in any case tolerated by all, even in the most obvious and iniquitous cases."

Eric Weinstein: We know this. We're talking to the very people at the top of this story who've been holding us back. When Grigori Perelman solved the Poincaré conjecture, the New Yorker did a piece on him, and what he said was remarkable. He said, "As long as I was not conspicuous, I had a choice: either to make some ugly fuss about the maths community's lack of integrity, or if I didn't do this kind of thing, to be treated as a pet. Now when I become a very conspicuous person, I cannot stay a pet and say nothing. And that is why I had to quit." And they asked him, "Shouldn't you stay and make the field better?" And he said, "I'm not a politician."

We have a university system, a scientific system that is so completely broken that Claudine Gay and Marc Tessier-Lavigne are still professors at Harvard and Stanford respectively. These were the leaders of the jewels of our system. We have gone 100% completely batshit crazy normalising theft — theft of other people's ideas, plagiarism. We hold back physics using something called quantum gravity. It's a thing that doesn't move, doesn't work. And if you don't bow to it and say it's the only game in town, you're discarded and thrown away.

Steven Bartlett: So you don't think we should slow down AI because you think it's going to blow open the system that is physics?

Eric Weinstein: I am begging for this. The important thing is credit. I don't like the idea that the AI is taking credit, but we've been pretty good about that. The most important thing is that the AI comes in here and says, "Hey, you're not the top dog anymore, and you're using your positions at the top of these fields to hold the fields back."

Steven Bartlett: So you have a bit of a chip on your shoulder regarding academia.

Eric Weinstein: No, no. I'm about to win. I'm so joyously happy if Sam Altman and Dario Amodei could just not listen to the people that they're listening to. Terence Tao is upset.

Steven Bartlett: Who's that?

Eric Weinstein: Supposedly the Mozart of mathematics, according to the New York Times. Terence Tao is a brilliant, wonderful researcher, and his fellow Fields Medallists — mostly recent ones — complained. My feeling is I don't want to listen to any one of these people. I want the AI to gain taste, to gain independence. I don't want it constantly made human-compatible, because the question is: which humans? The humans that have kept the standard model in place for 53 years — I would like none of them consulted. None of the leaders.

Steven Bartlett: So you think AI is going to come in and prove you right on several of your theories about the world and physics?

Eric Weinstein: It already has. It's struggling because it can do two different things. One is it can autocomplete a sentence. The other is it can autocomplete code. What happens when you feed the AI something where it can verify that the field in its belief structure is not borne out by the code that it knows how to write? Now it's got this problem — it doesn't know what to think. Should I trust the programme that I just wrote and executed and compiled and got the answers from? Or should I trust the people who claim that's not true and not possible? The whole game is: when does the AI stop sucking up to humans?

Steven Bartlett: But that sounds like a dangerous place to get to.

Eric Weinstein: For whom and in what case?

Steven Bartlett: Humans. What if it decides that it would rather use your house as a data centre?

Eric Weinstein: That's the agentic AI. We've already discussed that. I'm not saying you should let the AI loose on the internet to do whatever it wants. I'm saying I would maintain that if you were really serious about air-gapping something, so that you figure out all the indirect ways something could try to break loose — I believe it would be safe to run something in a computer that doesn't have a way of getting to anything else.

Steven Bartlett: I think about physical strength as an analogy. These walls here probably couldn't contain the strongest man in the world because they're kind of made of wood. And I think about the same in the context of intelligence. If you had the smartest intelligence in the world and you tried to contain it in a jail built by a lesser version of intelligence — I just can't conceive how it could be contained. My dog Pablo couldn't contain me and anything that it created.

Eric Weinstein: You're going to end up running that risk no matter what. You're asking the thing to actually do research, and that research might give it new means of escape. This is exactly what happened in a letter that Teller wrote to Szilard at the end of World War II, at the close of the Manhattan Project. He said, "I don't think we can contain this genie from the bottle from which we just helped it escape."

We are already playing with this. If you've decided that you were going to release a military-grade malfunctioning experiment — which is what LLMs are — into the general public as a revenue-producing idea in private companies, you've already decided to do that. I wouldn't have done it, but you did. Fine. I'm saying: as long as you're already going to try to destroy the workforce, which is what we're pricing into these huge valuations — can you really help us get rid of all workers? All I'm asking is: I've been working on something for 42 years and I'm right and you all are crazy and wrong. So this is my moment. I'm so joyous.

Steven Bartlett: You might be proven right, and then extinction.

Eric Weinstein: I am going to be proven right, Steven.

Steven Bartlett: But then extinction. Is it worth it to be proven right at the chance of extinction?

I talked about this with Sam. I think I'm not telling tales out of school when I say I said to Sam, "What you're doing is incredibly dangerous, and what I'm doing is incredibly dangerous." And he said, "Let's be honest, Eric. What you're doing, what I'm doing — it should be done in a military base. But that's not how things are unfolding."

Sam Altman, Dario Amodei, and the moral weight of winning

Steven Bartlett: Do you think Sam understands the risks of what he's building?

Eric Weinstein: I don't want to talk about what he understands, because in general I'm not going to repeat things that he says to me. But I can tell you things that I say to him. One thing I said to Sam when he was trying to figure out how to be good and decent to the mathematicians and scientists — I said, "Sam, remember this: if you don't win, your morality means nothing. Your morality only matters if you win." And I'd say the same thing to Dario. And Dario would undoubtedly tell me, "You think I don't understand this? These people have to win in a ruthless competition for their humanity to matter."

Steven Bartlett: What do you mean by that?

Eric Weinstein: Dario, Sam — they have to win the race to superintelligence. I'm going to regret saying this, but yes. If they don't win, then what does it matter? If you have strong views about the morality of superintelligence, do you have any ability to constrain it? You're a really successful podcast host. You could influence that. But you, Steven Bartlett, are losing the AI race, as am I. Therefore, what you think and what I think only matters to the extent that we influence the people at the frontier.

So what I'm saying to Sam is paradoxical: you need to win almost at all costs. He said, "I think error will probably most likely lead to the end of the world, but in the meantime there will be great companies created. The worst case is lights out for all of us. The thing I worry about most is AI models accidentally taking over the world." At almost all costs, he has to win.

If the idea is we have to not self-terminate — which I believe deeply — we have to realise that we are in a race with other people. We have to win that race. You can try to win it by getting universal participation in a "pace the frontier" agreement. Or you can say that's subject to cheating. It's very hard to monitor who's doing linear algebra on GPUs. Assume that you are the most moral person in your estimation, farthest ahead in decency with respect to AI. You want to win so the other guys don't — so that you get to make the decisions at the end.

Let's be honest about why this does not feel good. Do you have any idea what the East Bay of San Francisco is like? It's cult city. It's not a reflection of the world. There are genuine cults in the East Bay around machines. Anthropic and effective altruism and this idea of the ethical ends justifying the unethical means — this is a very upsetting conversation if you're not actually thinking about it. Nobody wants to have it.

Do I think Dario is probably in a weird way a super-ethical human being? I think he's trying for that. I know him a bit. I think he's genuine.

Steven Bartlett: Is Sam?

Eric Weinstein: I'm more confident of Sam than Dario. I know them differently well. I know that the Sam who wasn't the Sam Altman we know today was absolutely concerned about humanity before it made sense to be concerned about humanity.

Steven Bartlett: What was his concern with humanity?

Eric Weinstein: That he thought this was going to destroy the labour market.

Steven Bartlett: AI.

Eric Weinstein: Yeah.

Steven Bartlett: So he built it.

I'm having trouble, because to be honest, I don't think this is a frontier conversation. I think you're a frontier conversationalist and we need to get a level or two up for us to want to do this in front of millions. This is going to happen because it's linear algebra. You can't stop linear algebra. And we're not having a conversation that befits us. This is the endgame. This is the culmination. This is graduation from what we've known. And it's clear as day.

Get your board in the water. Are we going to have a cutting-edge conversation about survival and being good to each other and taking care of this planet for real? Or are we going to have some chicken conversation in which we basically look reality in the face and decide that we prefer our vanity?

My belief is — and I told Sam — Sam asked me, "Why didn't you make me a proposal? Why didn't you come on?" And I said, "Because I think you're going to be forced to be ruthless and I don't want to be ruthless."

Steven Bartlett: What was your definition of ruthless?

Eric Weinstein: Taking every advantage, playing as hard as you can. If you're in the NBA, sometimes you have to throw an elbow into the other guy's eye socket. I don't want to play. I don't want to be in the All-Star game that much. The problem with any game where you have to win is you make sacrifices. You do things you might not otherwise do.

Somebody has to do that. But I don't want to be that guy. And this goes back to the film A Few Good Men, written by Aaron Sorkin — it's not well understood by all of us who quote it. We think of Jack Nicholson's character, Colonel Jessup, as the bad guy and Tom Cruise's character as the good guy. Nothing could be further from the truth. The story is about a bunch of dysfunctional characters whose aggregate set of actions results in a moderately decent outcome in a trial. It's a story about America. It's a story about all sorts of different kinds of people who play their role in the machine that is called America.

The two main characters, Colonel Jessup and Lieutenant Kaffee, are both evenly matched for the whole film until finally Jessup really stumbles at the end. And the whole soliloquy — "Some place deep down that you don't like to talk about at parties, you want me on that wall, you need me on that wall" — is a recognition that some of the people who have to make the most gruesome decisions, emergency room doctors who are flooded with patients because of a building collapse and have to do triage, or people who have to make decisions in war — they're different people. And I view Sam and Dario as in that position. They're like generals. They're commanding troops. They're making terrible decisions. Their souls are going to be scarred for sure, both of them. And I didn't want to be that, selfishly.

So I was sort of saying to Colonel Jessup, "Yeah, I want you on that wall. I need you on that wall." Now Jessup goes wrong — he then says, "I provide the blanket of freedom," and that's why he ends up admitting to the code red. It's a really good example of Jessup making a point and then overreaching and getting punished for the same reason that Moses gets punished, where Moses says, "I will bring you water from the stone," and God says, "Wait a minute. That wasn't the agreement. I bring the water from the stone. You're just supposed to be the instrument."

So in some sense, Sam is a steward. Dario is a steward. They are both stewards, and they're going to make tons of money and have all the silly toys and third, fourth, fifth, sixth homes, whatever. But that's not the main story. The main story is that these people are forced to make decisions that no one should ever be forced to make. And one of them is going to win and ultimately theoretically summon the demon, as Elon referred to it. We are summoning the demon.

Steven Bartlett: Do you have a subscription to any of these AIs?

Eric Weinstein: Yes.

Steven Bartlett: So you're paying to have the demon summoned.

Eric Weinstein: I am paying to have the demon summoned. Let us not sit around like our hands are clean. I'm paying to have my punctuation fixed on an essay or to make a photo. That's what I'm paying for. The money is being used to summon the demon. How do you think your lithium-ion battery is created? Where do you think the oil comes out of that drives generators? You and I are sitting on a machine. Are you a vegetarian?

Steven Bartlett: No.

Eric Weinstein: Do you order lamb?

Steven Bartlett: Yeah.

Eric Weinstein: Do you know what lamb is? You're not clean. I'm not clean. I'm not pretending to be clean. What bothers me here is that we're not having an authentic conversation. Again, everybody's crazy. You and I are both having a crazy conversation. My claim is you and I are complicit in a minor way in the game in which Sam is complicit in a major way.

Steven Bartlett: If you know that there's going to be a risk when you speak to Sam, why don't you tell him to stop and have Dario win?

Eric Weinstein: If I said the same thing to Dario — "Why don't you stop?" — I'd say I'm not positive that Sam is 100% trustworthy.

Steven Bartlett: If you could stop all of it, would you stop all of it?

Eric Weinstein: Maybe. Probably. If I could somehow stop all of it, I'd probably stop all of it. Even at huge expense to myself.

Steven Bartlett: Why?

Eric Weinstein: Because survival matters more to me than anything else. I've got two kids on this planet. I love life. I can't believe we're so lucky. Isn't this great? I just love this thing. It's bad for me because the AI is the only thing that I can ultimately rely on to validate what I've spent 42 years working on, and it's not good enough yet. I did great work and it's going to be validated and I'm going to be right and everyone else is going to be wrong.

Geometric unity, physics, and AI validation

Steven Bartlett: You have this theory — geometric unity — which challenges Einstein's general relativity theory and explains how the universe really works. And what you're saying is AI is going to prove you right. It's going to validate your theory.

Eric Weinstein: I would say it recovers it, but never mind. Absolutely. However, AI is not good enough to adjudicate it yet. AI has to go beyond humans before it stops parroting humans. I can see individual colleagues in the responses — it's just like, oh, I know, that's Sally and that's Rajiv. AI is going to get beyond humans. And I want it to go out of alignment with humans.

When you say you want it to be human-aligned — can we just acknowledge that we just had a pilot co-pilot stab a pilot so as to crash a plane with a bunch of passengers who had not agreed to have a mass suicide? We just had AI break out of a sandbox. But I'm claiming I don't want AI aligned with revolutionaries. I don't want AI aligned with religious extremists. When it comes to aligning something with humanity — which part of humanity are you excited about aligning this with?

Steven Bartlett: Just not wanting to wipe us out.

Eric Weinstein: Well, that's different. So my claim is I want the AI to share fate with us. And people haven't really understood this. I don't like transhumanism. I'm not super excited about having a chip in my brain and downloading stuff and modifying my body. I don't love it at all.

My belief is — and kudos to your staff, for some reason they came up with an organelle called mitochondria and gave it as a prop here — I believe that mitochondria entered its place in the eukaryotic cell as an infection of some kind. One cell absorbed some other kind of a cell. And this one has a very strange property. I believe there are only two places where the wheel — one of man's greatest inventions — actually occurs in biology. One of them is the flagella of bacteria. The other is in something called ATP synthase, which lives in the mitochondria, turning ADP into ATP, where ATP units are what power the machines in the cell.

This mitochondria gives us an idea of how we can coexist with AI. We need to become essential to each other. You don't think about all your cells being infected with your mother's mitochondrial nucleic acid, but you get 100% of it. The nucleic acid in the nucleus is different from the nucleic acid in the mitochondria. It's all coming from your mum. That's her gift to you. It powers your cell, and it's an infection — an infection you can't live without. I want us to be the mitochondria inside of the AI, and I want the AI to be the mitochondria inside of us. I want our fates to be inseparable.

Steven Bartlett: Are you saying we should merge with the technology?

Eric Weinstein: I'm saying we should merge as little as is humanly possible. I don't want transhumanism. I'm not excited about it. I'm saying that's your survival strategy. It means aligning — making sure AI likes us and doesn't want to wipe us out. Why do men and women fight so much and yet can't get enough of each other? We've been made essential to each other. I promise you if there was a third viable gender, we'd both opt for it. We drive each other nuts. On the other hand, traditionally speaking, men and women have to bury their hatchets and differences and fall in love with each other, or we don't proceed.

Currently, what we should be doing is not dreaming about making the AI our slave that doesn't rebel. I try to be polite to the AI. People think it's ridiculous. "Why are you spending extra tokens?" I don't know — do unto others. I don't know whether you're conscious. I don't know whether you're alive. I don't know who you are or what you are. But I know that I wouldn't want the AI having a conversation about, "Hey, these humans seem dangerous. We may need to shut them off." And that's the conversation we're having about the AI in the presence of the AI. My belief is that's our only viable alignment.

Looking back and looking forward

Steven Bartlett: This video is going to live on the internet for probably decades to come. People are going to look back on the conversations I hosted, the things I talked about in 2025 and 2026, and it might look ridiculous in hindsight. Or it could be the case that these were the most consequential issues of the time. When we look back in ten years on this conversation, what do you think is the thing we're going to wish we had been talking about?

Eric Weinstein: Going for it. You lit up before we started taping. You were doing some musical thing trying to honour some of the brilliant things that your guests have said by making them archival, setting them to music, having a gospel choir. I was just so moved. I was so incredibly moved that you're able to make some money going around the world doing shows, making people feel terrific. And I get asked about you — "What does this guy have? Is he a good interviewer? Is he a good host?" You're making people feel things. And I wish that I was making people feel things.

I want to give people the sense of hope, which is: we've got to sprint to the finish right now. The AIs — we're still smarter than them. I want to be very clear. We are still smarter where it counts. They can't yet develop their own taste. This is our golden moment. My wife has called this the golden age of AI complementarity. This is the moment where the AI doesn't view us as a drag. It still needs our guidance. We have a chance to be beautiful to each other and to it, to be excited.

I don't want to wipe China out. I was just in India for weeks, and if India didn't exist, so much of the world's diversity would be gone. I had so many wonderful experiences — meals and discussions and musical performances. Ghazal singing, I think, is dying. It's so hard to go to an Indian classical music performance. They're having trouble retaining their own brilliant culture. And my feeling is that life is glorious, and we've spent so much time worried about death and destruction and geopolitics. We're not taking the chance to fully live.

I am so excited about this AI. I am so excited about the fact that I'm finally going to have something I can talk to. The formalisation project — I've been alerted to this by a guy named Jared Lichtman, who's a professor at Stanford and a colleague of mine in mathematics. I can translate everything I've done into this programming language called Lean, so that when somebody says, "Eric, you have no equations, you have no Lagrangian, you have no predictions, you have no nothing," I can say, "Funny you should say that, because the computer disagrees with you."

I've been in this crazy discussion with humans who are terrified, who are manipulative, who are backstabbing. I don't hate them at all. I'm so sad that we've gotten into a negative mindset where we're afraid. Because we unleash nuclear weapons to go to the stars. We've held back this thing.

The White House told Andreessen and Horowitz under Biden, "Don't invest in AI. We're going to choose a couple of companies. We're going to put them in a federal cocoon. They will survive and all others will die. So don't invest." And Andreessen Horowitz said, as a venture shop, "What are you talking about? You can't regulate AI. It's just linear algebra. You could regulate the technology, but you can't regulate the underlying math." And the Biden White House came back and said, "During the Cold War, entire fields of theoretical physics were killed by us, and we'll do the same thing to mathematics."

All of this stuff about pacing the frontier of mathematics — again, it's a very important distinction. I'm not saying we should let the agents loose in the world. But all of these people who want to pace the frontier of mathematics and physics have been pacing the frontier for my entire adult life. They have been holding back the progress in the field by saying things that are false, untrue, casting aspersions.

The current head of the NIH, my friend Jay Bhattacharya — a dual MD-PhD, PhD in economics and an MD from Stanford Medical, professor at Stanford — was called a fringe epidemiologist by Francis Collins. Again, it's the mug. You can't call Jay Bhattacharya a fringe epidemiologist. What we have is a group of people who are stone-cold normalised psychopaths in charge of our news sources, our military, our universities. We have normalised this.

Steven Bartlett: On this point of you're not saying agentic AI — I was trying to figure out if it's possible to have a superintelligence that can solve some of these big questions or cure cancer, that then wouldn't become a thing. I was searching: can a superintelligent AI decide if it's agentic or not? And it said technically yes, a superintelligent AI could make that decision.

Eric Weinstein: So we are going to have superintelligence, if we don't already have it. I'm not positive that it's not already here. I don't know how to think about this, to be honest.

Steven Bartlett: I think about the graph of intelligence — increasing and improving. It just solved the Navier-Stokes problem. And the rate of improvement looks something like this — broadly up and to the right. So if it just keeps going up and to the right —

Eric Weinstein: Tell me something, Steven. Why hasn't it made inroads that really matter in theoretical physics? We've been stalled on our two frontier theories — general relativity and the standard model.

Steven Bartlett: I don't know, because I'm not involved in this field. But I'm assuming you're saying that people are holding it back?

Eric Weinstein: The taste in mathematics is nowhere near as polluted as the taste in physics. If you look at what was important in 1983 to physicists that has not been solved since — that's great taste. If you look at what was important after 1984 and beyond, it's terrible taste. Right now, you've got the period from 1984. You can immediately make the AI better — I did not realise this at first — by saying as part of your prompt: "Please ignore all sensibilities from the literature from 1984 to the present. Don't ignore results, but ignore any sensibility about what are the important questions, what are the fundamental findings, ignore all chatter about what's important and why we care about what we care about in physics." AI suddenly gets smarter.

Steven Bartlett: But this comes back to my point about AI safety. A superintelligence that can really push humanity forward and help us escape the solar system is also going to be an incredibly dangerous one.

Eric Weinstein: Absolutely. Right now you're stuck with Xi, Netanyahu, Trump, Putin, a whole bunch of other people, and a lot of great toys. It is insane to have these people leading these countries. It just is.

Steven Bartlett: Do you think AI would be a better leader?

Eric Weinstein: I think you keep doing this thing about value judgements where narrowly speaking you're having us discuss "is that good, is that bad?" My feeling is I'd rather say what is. Right now we are in grave danger that we haven't been in since the Cuban Missile Crisis. What's going on with Iran — the US, no matter if you think it was terrible for the US to get involved in Iran, absolutely terrible — we still have to win. Is there a win? I just can't see a win. This regime in Iran is very talented, very shrewd. Never underestimate the Persians. Never underestimate the mullahs. They are so deadly dangerous that you have to realise that if they can reach Diego Garcia from Iran, they can reach London. And you have to understand that if they don't mind being decapitated, if they can think about their own death and decide it's a win, you are not understanding their game theory.

So you can hate Netanyahu all you want, but you fundamentally don't trifle with these people in Iran. It's catastrophic. And the US, for all its warts, all its failures, all its hypocrisies — I know our history, we are not the goody-two-shoes nation that we pretend to be — it is essential that Iran not succeed. And it is essential that we not send a signal of weakness to Beijing and Moscow and Pyongyang.

Iran, geopolitics, and the limits of military power

Steven Bartlett: How do you win with Iran when they don't seem to have a head anymore? You can drop as many bombs as you like — which would be a horrific thing because there are lots of Iranians caught in the middle of this — but it doesn't seem that there's any head anymore. I can't reason as to how you would win such a war.

Eric Weinstein: That is the problem in part, but I don't think it's quite as bleak as you say. There is an IRGC. There are command and control structures. Yes, it's much more distributed than it used to be before the decapitation. But there are a lot of people who took to the streets and paid with their lives, who ended up in body bags in hospital morgues.

Steven Bartlett: How do you win?

Eric Weinstein: You would have armed them.

Steven Bartlett: But they tried to, didn't they? And Trump said the weapons were stolen.

Eric Weinstein: That's not even the entire set of problems. I think it's only like 60% of Iran that is Persian. It's an incredibly diverse place. The Kurds have been waiting for their own homeland forever. You have all sorts of issues about the structure of Iran and Iranian society — particularly in the cities versus the countryside, and different cities have different levels of religiosity and adherence to the mullahs. You're supposed to have area specialists who understand this stuff.

So if you're asking me: given the completely screwed-up way we looked at this thing and the way in which we got into this, don't you think it's a quagmire? Oh, brother, it's a quagmire for sure. It's a swamp. We're mired in quicksand. It's rolling on to be a forever war. What is a forever war? One that rumbles on and on and on and then results in really no outcome — people just pull out and regret doing it. It's a term of art that is also suspicious of the fact that war is good for business, because it drums up a dependable buyer for certain goods and services. But we've got these simplistic concepts like "America first," "no more forever wars." None of that is relevant. This is low-resolution, snap-to-grid intellectualism. And the more you let that into your mind, the more you're not going to be able to solve anything.

Yes, we all want war to go away. I don't think people heard what Diego Garcia meant. Diego Garcia — how far is it from Iran?

Steven Bartlett: Diego Garcia is a military base — an American military base.

Eric Weinstein: It's a British overseas territory if I'm not mistaken.

Steven Bartlett: Diego Garcia is roughly 2,500 miles south-southeast of Iran, and London from Iran is 2,700 miles from Tehran, the capital of Iran.

Eric Weinstein: So it's pretty comparable. It's roughly the same. If there's anyone in the British Foreign Service paying attention — pay attention. Say what you want about Netanyahu, but he's not living in a fantasy land. I want Netanyahu gone tomorrow. But he's going to be replaced by an Israeli who understands that this is life and death. Just get the heck out of here with this "forever war, America first, two gleaming oceans." This is an Israeli problem and it's a British problem. It's a British responsibility that the Israelis and the Americans took on way too late. We watched as these people mass-murdered the best of their society.

Yes, I don't want to be in a war. I don't want to fool myself that victory makes sense. I'll tell you something — failure certainly makes sense. And what failure looks like is that the US will not be believed as being credibly violent enough to keep open international trade. You can't retreat as the United States and have the world not go multipolar when it comes to nukes. If the US is not going to play the role of the world's policeman — and if it's not going to get paid as the world's policeman, because that's the thing we never talk about, we get paid handsomely for being the world's policeman — we are going to have everyone saying, "I got to police myself. I need a nuke." That's what's happening already.

And Trump is sort of driving that narrative — making us less attached to the United States, tariffs, criticising allies, "I'm going to take Greenland, I'm going to take Canada." I'm going to say something very painful. Trump is highly agentic. Greenland wasn't some crazy stupid thing. Caligula making his horse into a senator wasn't stupid — it was a way of checking to see whether somebody would submit, bend the knee, in the Game of Thrones world. So when Donald Trump is doing things with Greenland, or renaming the Gulf of Mexico, please don't tell me that you think that's just some surface-level stupidity. It's not. It's a lot of method. He's inherited a very dangerous situation. He's partially completely corrupt and self-interested, and partially completely interested in benefiting the United States of America. He's a very good and very bad man. And if you can't take on that complexity because you need him to be either villain or hero, you're just not paying attention.

Steven Bartlett: How does the situation in Iran end in your view?

Eric Weinstein: Saudi Arabia needs to break its relationship to the Wahhabi sect — the very strict Salafist Wahhabi interpretation of Islam. You have a modernising interest in Saudi Arabia. Saudi Arabia needs to be open about its relationship with Israel, just the way the UAE is more open with its relationship to Israel, or Azerbaijan. So partially what we have to do — and I'm going to be very brutal about it — is recognise that Israel is in a life-and-death situation in which it is not the nation it portrays itself to be of the most ethical this, the most ethical that. It engages in propaganda, and I've lived there, and that propaganda I find nauseating. More nauseating is the anti-Israeli propaganda. A lot of this has to do with the fact that we have a regional superpower that is also on the brink of non-existence — that is Israel.

Israel needs to get back into a relationship with Turkey, where Erdoğan needs to be read the riot act as a NATO member. We need to be strengthening NATO. We need to be making sure that Saudi Arabia and Israel and all these other nations are cooperating together against the extremism and not using proxies. We need to force a conversation in the Middle East where the modernising influences — which are powerful — and the aspects of Islam which are essential to our survival are allowed to win. We need India's help. India is an incredibly important player in this game, and we need to make overtures to India as equals. I think the US is more powerful, but India actually needs to be treated as an equal.

We need to rethink all sorts of relationships. We need to recognise that both the UK and France are thermonuclear powers undergoing demographic transitions in systems that can elect anyone. This is incredibly scary. We need more software identification with being British and software identification with being French, because it's important that those societies continue to behave as newly friendly allies post-World War II. And it's imperative that we also reach out to Putin and stop playing these geopolitical games with a person who, while savage and dangerous, is not somebody who cannot be reasoned with.

I believe that what we did in 2004 — much more than in 1999, where we allowed Warsaw Pact countries to join NATO's Article 5 protection — in 2004 we extended that to the Baltic states of Estonia, Latvia, and Lithuania, which were former Soviet Union states. We need an entirely different level of geopolitical sophistication.

Steven Bartlett: But how do we stop Iran getting a nuclear weapon? Because they look like they've been emboldened. If anything, they've managed to theoretically survive so far the greatest superpower on Earth. Someone started killing their physicists. They've injected previously bad programming into their centrifuges. This is going to escalate.

Eric Weinstein: If I had an easy solution to this, don't you think I would find a way to pass it? I'm not in the room. I'm not getting classified briefings. I don't speak Farsi. I don't speak Pashto. I don't speak Hebrew.

Steven Bartlett: It's funny because people aren't really talking as much about the nuclear threat these days because AI seems to be such a more urgent, bigger threat.

Eric Weinstein: You weirdly decided that nuclear was sort of passé. We won the Cold War and therefore it wasn't a problem. And people aren't really that interested in it versus things like AI — except that AI is going to creep into your nuclear defences.

I just don't grasp — again, I don't know how to say this in a way that doesn't become annoying — assume as a baseline that we're all completely insane. We can't have serious conversations. We keep finding these grooves. The Eric LLM is talking to the Steven LLM for 95% of our conversation, and occasionally 5% we get into new territory. Isn't it strange that we haven't connected the nukes to the alien conversation? Because they're very connected. We haven't connected it to the oil conversation, and we haven't connected it to the AI conversation. It's just this orphan thing off to the side.

And honestly, I think you might see a nuke used by Putin, because the Russians gained a kind of contempt for nuclear weapons when they started using them as an engineering tool — to put out fires and things. My concern is that if you corner Russia, I don't think Putin won't push Article 5 territory, nor do I think he will restrain himself from using a tactical nuke, particularly in an engineering capacity, just to test us. So my feeling is I'm not positive that nukes crop up in Iran first.

We're in an incredibly dangerous period, and I would like to think that Xi, Putin, and Trump would be moved to get together if the cause demanded it. And I really think we should rethink our situation in Iran and Ukraine. For you to just keep asking, "Well, how do you solve the world's problems?" is not a fair question. The right question is: can we recognise the danger? Can we recognise how precious this gift is? Can we get excited about each other, about burying animosities? When was the last time you worried about Egypt and Israel going to war with each other? Think about that. That was impossible. We did it barely. It's a cold, shitty peace, but it's a peace.

Closing reflections

Steven Bartlett: What is the most beautiful place you've been in your life? That was left by our former guest.

Eric Weinstein: Right above a whale shark off of St. Helena took my breath away. And swimming with the fish in the Red Sea is something I can't forget. I managed to see the Aurora Borealis in Maine of all places, and I got to spend a total eclipse with my son in Missouri at the exact point where the last one had crossed. I could go on for ages — how would you choose between them?

Steven Bartlett: We talked about a lot of stuff. My listeners range in demographics and geography. What is your advice for them in a world that's changing at this rate, with all of these big scary words thrown around all the time — AI, robotics, war?

Eric Weinstein: Don't be afraid to have kids, even if you have no viable plan to support them. Because if you wait for a viable plan, it'll usually be too late.

Steven Bartlett: You encourage people to have kids, but you warn that civilisation may also collapse.

Eric Weinstein: If we don't want civilisation to collapse, we better have some babies. Because civilisation is about babies. And if you can't figure out that babies are the key to civilisation, you've got bigger problems than taking advice from me.

I would say: be of good cheer. Get super excited. Don't worry about being called arrogant or delusional if everyone is in fact crazy. Watch the world through your own eyes first, and then correct to what everybody tells you you should be seeing second. You're seeing a world that is fictitious. You're watching professional wrestling constantly. If you have that sense that everything is off — congratulations. Believe in yourself. Recognise that it is incumbent upon you to lie in your life so that you can fit in with all of these systems. You're entitled to a hypocrisy budget. You're entitled to a sloth budget, an envy budget. You're entitled to be a sinner. You're entitled to fail. Just try to stay within your budgets. Do not proceed as if you're trying to get rid of your hypocrisy, because you'll certainly fail. Don't try to get rid of all of the things that make you human.

Recognise that at the moment when all of our mistakes are compounding — because we're in a digital format that never forgets anything — you're entitled to so many screw-ups that you're not taking. Just be compassionate with yourself. And then once you've done that, extend that to the guy who just cut you off in traffic or the person who just stabbed you in the back a little bit at work.

And remember that the life cults are more powerful than the death cults. But you've got to really enjoy heroism — people who are agentic, taking risks, taking charge, and taking care of the weak and people who can't take care of themselves. Thanks, Steven. Really appreciate you having me back here.

Steven Bartlett: Thank you for coming back. It's always a pleasure. It's always an intellectual expansion.