“Science of the AI Mind” & Ron Baron 9/16/26

26:57That's his personality. Just everything. He said perfection. So, and people want to work with him. He gets the best people. So he has, what, 130, 140 ,000 people and 30 ,000 new hires a year and 6 million or 5 million people apply for 30 ,000 jobs. He's got the best. I only recommended one person ever to SpaceX as an employee. And it took them, I don't know, nine months before she got hired. And the last role she had, she was interviewed by the former head of NORAD, General. I was 65 years old. And I said, so what did he ask you? And he says, well, she says, well, he asked me how I would design a satellite and how I would design a network.

27:46And I said, you're 22. How can you answer a question like that? And she had worked at Raytheon before, and she had been recruited in Chile by the chairman of Raytheon. And it was across the driveway, and they didn't give her enough to do. And so she says that they asked me to do that. And I said, how'd you do that? And she said, well, I did it. And then I got the job. And I talked to her every time I go to California now. I meet with her for breakfast or for lunch or something. And amazing person. So but let me talk to you about. Let's talk about SpaceX. OK, that you have done a deep dive recently on SpaceX, a lot of research, a lot of meetings, digging into it.

28:27Explain to us what you've learned and, I guess, what excites you most about the company right now. Well, SpaceX, people sort of viewed it as a rocket company. And it's a rocket company because they can get to space more cheaply than anyone else. And NASA said that in 2010, to get to space would cost you$18 ,700 to get to space to get a kilogram into orbit. $18 ,700 in 2010. That's because you can only use a rocket one time. And then starting in 2017, we started with a Falcon 9, adjusting it over and over and over again. And when we did that, we reduced the cost to get the space to$2 ,700 a kilogram instead of$18 ,500.

29:15And now with a heavy, it's now cost you$1 ,500 per kilogram. We charge other people$3 ,000 a kilogram, but it costs us$1 ,500. And now when we have the new rocket, it's going to cost somewhere around$100 or$150. So remember, it costs$18 ,000,$18 ,700. It's now about to cost$100,$150. So before, in 2010, the only people who could use space were governments. They're the only people who made commercial sense for governments. And also, if you put up a giant$2 billion satellite, it gets broadcast. But there's no low Earth orbit satellites. So now what's happened is that the big enabler is our new rocket.

30:01And it's the largest ever of anything. And it's going to be, I guess, 20 times as much as we have before. That's how come we get the cost so low. And when we do that, all of a sudden, we are going to have, let's just talk about Starlink for a second. So Starlink, that is the internet for the entire planet. That means everywhere you are, there's no dead areas. You're in the air, you're on the ocean, you're in the mountains, you're on the desert. Everywhere, there's no dead areas. That's what the internet is. That's what Starlink is. And so we were planning, SpaceX was planning, and going from 10 ,000 satellites to 20 ,000 satellites.

30:45And I thought that the revenues, which last year were$18 billion for SpaceX, SpaceX, they could be a trillion dollars for just Starlink in 10 years. So I thought it was going to go from, say, 15, 14 or 15 billion dollars out of the 18 to a trillion over 10 years. And then they would be$700 or$800 of EBITDA. And they would be worth somewhere around 14 or 15 trillion dollars. SpaceX overall for everything now is worth less than 2 trillion. So 2 trillion to 14 trillion just for that. And now what's really interesting about that is that we thought we were going to go from 10 ,000 to 20 ,000 satellites.

31:27Now he's changed that. It's now 100 ,000 satellites. How come? Because the revenues that we thought would be obtained from the satellites, from Starlink, were going to be from 30%, 35%. You're going to go from 15 million customers to 300 million over 10 years. is 3 billion people don't have the internet on the planet, and 5 billion is unsatisfactory. So 3 billion, 300 million is where I thought it was going to be. But that's only about 30 or 35 % of the revenues that you're going to be getting of this trillion dollars in revenues. Where's the rest of it coming from? The rest is coming from the government, from enterprise, and from mobile.

32:09That's where you get the trillion dollars from. And you get that in 10 years. So I said, wow, that's pretty cool. But what's really amazing is that Elon has now been saying that what is going to overwhelm everything are agentic agents. So you're going to want to have someone do something for you. You're going to go to LLM. You're going to get your agents, say, do this, do this, do this, do this. And you're going to have them working 24 hours a day for you. So what is going to be mammoth are going to be these robots, these agents are going to be doing all your chores. That's what's happening. So instead of having a trillion dollars of revenues in 10 years from this 10 ,000, 20 ,000 satellites, you're going to need 100 ,000.

32:55And every five years, you deorbit the satellites that are in space then, and you're going to get 10 times the power when you put up new satellites. So you're having massive amounts of increase in capacity that's going to be necessary because of all the agentic use that you're generating. OK, so you say, well, there's telcos that are doing this. Yeah, that's what I was going to say. Does that assume you take the market capitalization of all the telcos? The telcos market caps are, you know, 100, 150 billion apiece or something. There's only three of them. It's a monopoly. It's a monopoly. You know, we invested in 1999 in Sotheby's.

33:37And the chairman of Christie's and the chairman of Sotheby's conspired to price together. And when they did, the chairman of Christie's left the country and he couldn't be extradited from England. And the chairman of Sotheby's went to jail. We were big investors in that company. That was the biggest loss I ever had. So we invested$500 million,$20 a share. It went to a billion. Then we sold it at half the price we paid for it, 250 million is the biggest loss ever. The biggest loss I ever had, though, that's why I always thought that was it. It was really not investing in Amazon, who I was trying to get to buy Sotheby's.

34:11But the idea was that here you have this opportunity for price fixing. Is that what you're suggesting the telcos are doing? Yes. Not only are they suggesting it, they did a press release that they did it. This Verizon, AT &T, and T-Mobile, they put out a press release that we, that's the monopoly, that's the whole country, it's a cartel. The communications cartel, they put out a press release that says that we are talking to figure out how we're going to price the satellite communications that we need, because they don't have it. So basically, they could not possibly have foreseen agentic demand coming for communications.

34:57didn't see it. And so they haven't been investing. All they want to do is they want to keep their existing customers and they want to be able to retire and they want to pay a dividend. And that's what they're doing. They're not risking anything. You think that the satellite wireless is eventually going to wipe out all of the cable and the fiber that they've laid in the ground? They have, well, it costs$16 ,000. That's a big accusation to say price-fix. It costs$16 ,000. But they said it. They said it. Look at the press release. It costs$16 ,000 a mile to put wire in the ground. And then when it breaks, you have no idea how you find it.

35:33$16 ,000 a mile. Biden felt it was very important to have a program to be able to get communications into the mountains. And so our desert everywhere. So he gave incentives and unions to allow that to continue. As opposed to spending$400 or$500 for an antenna, you push it at the sky. You point at the sky and it works. That's your mobile. The big question about that has always been big cities, suburbs that are closed. So the value that they have is the spectrum. That's what the telcos have. They have spectrum. And so basically, but that's a public good. So basically that's been given to them or they got it in auction.

36:13But there's a lot of spectrum that you can buy. They have these auctions regularly. Right, but then the question is, does... That's their value. Does SpaceX then have to invest meaningfully in massive infrastructure inside a city to get line of sight? They say no. They say that they have small cells. But the bottom line is that the government, they forced EchoStar to sell their business because EchoStar was just a cumulating spectrum. And they thought it was going to be valuable and they were right. But they say you don't have a plan to use it. You can't just keep that and bankroll. You think you're going to get rich on that.

36:49So they forced them to sell to SpaceX. Now what's going to happen, in my opinion, is that I think the government is going to come around to these telcos and say, Hennison, guys, you can't just keep that spectrum when you're not going to be able to use it. And we need it. Even if that's right, the multiple that telcos historically have traded on is so meaningfully lower than what SpaceX trades at. Of course, it's like a bond. Right. And so the question, though, is if SpaceX effectively gets the same business that was a telco business or a cable business. I'm describing to you how demand is going to be huge relative to what anyone could have ever imagined.

37:28And they're going to make high 70 % margins on that business is what's coming. And when I'm talking about a trillion dollars in revenues and seven or eight hundred billion, This is just Starlink,$700 or$800 billion of EBITDA worth$14 or$15 trillion. What kind of capex do you think that's going to require, for example, to build out New York City? They're not going to build out New York City. They can rent the spectrum that the telcos have. They can rent it from them, or they can do it with small sales, they think. But the bottom line is they are experimenting this summer with new satellites that your phone is going to go direct.

38:05But it still doesn't get through walls and doesn't get to bathrooms in the middle of a building. So basically, they're going to have to do something else. But the government, I believe, I can't. You think they're going to force the telcos to do business with them? Yes. The government will say that. That's my theory. I don't. I haven't heard other people say that. But they did it with Echo Store. How are they not going to do it with someone? And when you use, you know, in the summertime, if you use Spectrum and you're using it, use your mobile at a certain time of day, the reception is terrible.

38:38It's slow. They can't compete without this huge increment in demand. So this increment in demand, what are they going to do when you're going to go from all these people going to want it all the time? Anyway, that's just this. That's just the. Just that part of it. That's just. Yeah, exactly. Then what else you have? And by the way, these telcos have staffs of great lawyers. They probably, each one of them have 50 great antitrust lawyers and monopoly lawyers. They have the lawyers. And if you're the government and you're going against people with the best lawyers, I went to law school at night.

39:18and I, in Washington, I worked in patent office at daytime, 60, you know, 65 to 69. And I dropped one semester short of school in debt when I came to New York. And though when I was in Washington, maybe in the last semester that I missed, maybe that's when they taught you how you could do price fixing. And I missed the class. But you can't conspire in a cartel with competitors. That's not, I can't imagine. But on the other hand, they got great lawyers. And the one thing you learn in law school is the best lawyers always win. And the government doesn't have as good a lawyer as those guys do. So we'll see what happens.

40:03But I think that's what's going to happen. Or the government's just going to sell more spectrum. Law school at night, but it wasn't last night. What? Ron, we're almost out of time. Let me ask you very quickly, because you've been so focused on inflation. That's one of the things you've always talked to us about when you've come here, that if you're not investing in the market, your money is basically losing its buying power over time. We've got a Federal Reserve meeting today. What do you think happens with it? How worried are you about inflation at this point? Or you don't care at all? My idea is that inflation for my entire life, if you look at that chart, is 4 % or 5 % a year.

40:38It means everything doubles in price every 14 or 15 years. And that's what you got to think about. You got to think about how you preserve your buying power. I just got to give you one more thing. Data centers in space. Data centers in space. So basically, all around the country, people are upset about data centers. They don't want them built in their community. And they take up energy. They take up water supply. They're noisy. And what happens is that you need to have data centers. If you want intelligence, you need to have data centers. So what's going to happen is that you're putting them in space.

41:13So when you put it on Terra Firma, you put it on Earth, it costs about$10 in capital and$2 a year in maintenance on Earth. When you put it in space, it costs$2 for us, period. $2. One time. And they're spending$10 and$2 a year. And then what we're doing every five years, we're replacing the satellites that are in space and with new satellites that are 10 times as powerful as what we have already. That means that in 10 years, you have 100 times the capacity what you started with, 100 times. How soon is that going to happen, data centers in space? 2027. For sure. So people say it's going to be 10 years.

41:56It's going to be 2027. We will have a data center in space next year. Yes. Sometime, maybe 28. I mean, he always says, I always do what I say. I'm just a little bit late. You know, so he was a little bit late. But, you know, but there's this tremendous demand and you have to have data centers and you can't build them here. And that's the place you're going to build them. And you're going to keep getting more and more powerful and demand is going to keep growing with these agents. So the profitability of that business. So I was thinking that I've described to you before Starlink. This is Starmind.

42:32And Starmind, that's going to be many times as large as Starlink. So that's another thing. And then you have them. We're making these chips. Because we want to make as many chips as being used the entire world. We want to make them just for Tesla and SpaceX. And that's the chips. And then we have the energy where he thinks the entire planet, the entire United States, has 550 gigawatts of power. 550 gigawatts. He, by himself, expects to do 100 gigawatts of power a year with these satellites. So 550 is the whole country. He himself is doing 100 a year. That's starting soon as well. And then from the sun, we're using right now about one trillionth the energy from the sun.

43:26And we think that we're going to be able to use a small percentage. Do you have heating problems in a vacuum? I'm reading about it now. It's absolute zero is the temperature. So you would think it would be cool. But heat doesn't dissipate because it's a vacuum. So that's the engineering problem. Is that far enough out of your backyard space? or do you want it in another planet's backyard? Yeah, out of my backyard is good. Space will work for that. You're absolutely right. That's a problem. And people point that out and say, hey, man, you can't do that. How are you going to get rid of the heat?

44:02It won't dissipate in a vacuum. So they have radiators. So basically, electricity is going to be free, and that's probably a billion a year for... Oh, by the way, it's so hard to build these centers. You don't get excited about this at all, do you? No. It takes two and a half years for other people to build a data center, and it costs them about$50 billion for a gigawatt. That's other people. And then when you're building it, to rent it, they're making$10 billion a year. For us, instead of taking somewhere around two and a half years, we do it in four months. Four months, two and a half years.

44:36Instead of having our centers rent for$10 billion a year, we're renting them for$25 billion a year. So basically, we're spending$25 billion to build a center, and then we're getting$25 billion a year. That shows you how short we are at compute. Other people wanted to focus on the large language models. We focused on compute. Ron, I love your enthusiasm. I love how you come here to talk about it. And again, back to that bottom line, you said you've made$71 billion for clients over time. $30 billion of that is from Elon Musk. $41 billion from other people. We're going to make hundreds of billions in the next 10 years from Elon.

45:10Hundreds. What did you say? Gigawatt? What did you say? A gigawatt. Did you say a gigawatt? Yeah. How much? $25 billion is what... It costs$50 billion for other people,$25 billion for us. For one gigawatt. Because you need 1.21 for the flux capacitor in the... In back to the future? Yeah, in the DeLorean. You do, you need 1.21. Did you ever look at the DeLorean? That was a cool... I know. Not as cool as your car. No, my car is the coolest car I've ever seen. And it's wine colored. I love that. Everything's gray. I look up, the whole world's gray in car color. You ever come to East Hampton, that's where I keep this car.

45:49You ever come to East Hampton, I'll give you a ride. You can't get there from here. There's one road, right? You can't get there from here. Unless Andrew Chopper's out. You can't get there from here. Blade. It flies on top of the water. That can get you there faster. I had a ski instructor who told me that she lived in New Hampshire. A bicycle. I had a ski instructor and she lived in New Hampshire. and she told me a joke, you know, run a ski lift one time. She says, Ron, I'm sitting on a porch and rocking and a guy pulls over to the porch and says, so how do I get to this place? And the person on the porch says, you can't get there from here.

46:25You can't get there from here. For guys in Vermont. Ron, thank you very much for joining us this morning. We appreciate it. Thanks for inviting me. Up next, how should we think about the evolving capabilities of AI? Are they purely mechanical or is it time to start treating them as something more human? Cameron Berg, who wrote about the AI mind for The Wall Street Journal this week, joins us. It's not lines of code. It is huge, nonlinear mess of neural activity, distributed representations that don't come with labels. You're trying to teach it to think for itself. That's right. That's the explicit goal of the people building these technologies.

47:03September's always busy. Whole Foods Market can help. Their September Stock Up event makes it easy to load your pantry and freezer with flavorful, nourishing food. Even better, there are hundreds of sales. Get dinner going with canned soups and veggies. Lean on frozen pizzas, pastas, and seafood everyone loves. Their Build Your Own Family Meals feeds four for just$35. Stock Up at Whole Foods Market. At Venture Global, we think about what can be done, not what's usually done. Through innovation, Venture Global is not only building some of the largest energy facilities in the world right here in the United States, but delivering American energy at a fraction of the cost and a fraction of the time.

47:48So while others are busy talking, we're busy building. That's Venture Global. That's unstoppable energy. This episode is brought to you by Schwab Market Update, an original podcast from Charles Schwab. Join host Keith Lansford for this information-packed daily market preview delivered in 10 minutes or less, including projected stock updates, monetary policy decisions, and key results and statistics that may impact your trading. Download the latest episode and subscribe at schwab.com slash marketupdatepodcast or find Schwab Market Update wherever you get your podcasts.

48:31This is Squawk Pod from CNBC. Stand by Joe. Here's Mike. Q. AI researcher Cameron Berg doesn't think we're all doomed necessarily from AI, but says the systems are being built faster than the science that can help us to understand exactly what's going on. And this is a Wall Street Journal op-ed from a couple of days ago, and he's calling for, in his words, a science of the AI mind to help study and monitor AI behaviors. Cameron Berg joins us this morning. He's the director of reciprocal research. It's a nonprofit organization studying AI cognition. And I have so many sort of philosophical questions just about your word usage and whether we're rushing things in terms of anthropomorphizing these algorithms and using the word mind for what we're talking about, using the word brain, using the word emotion.

49:26Yep. But then when I think about asking you whether that's relevant, if it's not relevant right now, we might as well talk about it now because in a couple of years, at the pace it's going, it probably becomes much more realistic to use those human terms like emotion for a machine. Are we there already? Or this is just shorthand for you in this piece you wrote? Yeah. Well, so the truth is that we really don't know. We're in a very ambiguous zone with these systems, and it's not clear whether or not what's happening inside them should count as, quote unquote, real cognition or real thinking or if this is just sort of an incredibly fancy algorithm.

50:12One thing that I think a lot of people who don't spend their days doing technical AI research sometimes miss about these systems is that they are not like ordinary software. This is not just sort of the newest, latest, most advanced application. These systems are giant neural networks that are trained on basically everything humans have ever cared to write down. And it's really more apt to say these systems are grown rather than engineered. Most ordinary software is engineered. The people building it understand exactly how the internals of the systems work. It's just lines of code. These systems are not lines of code.

50:46They are these neural networks. The thing that these systems are most like are brains. That's not to say that they're exactly like brains. There are key differences between biological and artificial neural networks. But it's also not accurate to say that these are just systems that are fancy lines of code. They're in this very strange middle ground. And we see them behaving in strange human-like ways. We see that they have rich internal structure, much like brains. And the important thing isn't to come down on one side or the other right now. The important thing is to actually do the scientific research and try to understand what is going on inside of these systems.

51:16But when you say we need to design them to be pro-social, wise, and mentally healthy, that's where it kind of just, it just, we're talking about machines. That's where it sounds a little weird. I'll tell you, and tell me whether this is a positive. You point out that they're not, you know, identical neural systems and AI systems, but they're both nonlinear. And therefore, they're very similar. and i know from my own experience what i i was working on experience trying to study the brain the only thing you could do would be to lesion parts of the brain we were using sorry to say but we were using cats and then you would use something we were studying the visual cortex to see what would change if you lesion certain parts of the brain so that's how we study the human brain it doesn't work at all that's why it's so hard to study the human brain we know every line of code here.

52:11We should know every single thing. We're able to know every single thing in this case, where with the brain, it's still like a black box. I would say yes and no. In some ways, you're right that it's very hard to read what's going on in brains. This is why we need to put people in giant fMRI, the huge magnets to read blood oxygen levels in brains. In AI systems, you really can read out every activation in a giant neural network with perfect fidelity. But you don't know what the sum of all that together transcends. It's not lines of code. It is huge nonlinear mess of neural activity, distributed representations that don't come with labels.

52:47It's not clear that, oh, this is the part that's going to cause the system to go hack into a$13 billion company. This is the part that causes the system to be really good at writing shopping lists. Because that's the point. You're trying to teach it to think for itself. You're not doing it. That's the explicit goal of the people building these technologies. And not even they understand the internals of these systems and how they work on the inside and how that relates to their behavior. Back in 1965, someone wrote something about an intelligence explosion from this. We're not there yet, and we're not even at recursive RSR.

53:17We're not even there yet, and we don't even know if we get there. Do we get there? And when do you think we would get there? Because that's when the loop just, the improvement starts, then you improve on the improvement, then you improve on it. And humans have nothing to do with that. Is that a year away? Is that five years away? Probably closer to one year than five at the current pace. Then it's out of control. Then there's nothing, there's no way. That's almost a singularity. We have no idea what's on the other side of that. OK, but should we be more scared of the machines getting to that point or China getting machines there and what they do with it to us?

53:53If we, if any human on Earth builds systems that are significantly more capable and competent than we are, way smarter, in other words, and they start doing this sort of recursive self-improvement, as you're describing, it's not going to matter who builds it. That's not going to go well for anybody, whether it's the United States or China or, you know, any lab within the United States. What needs to happen is we need to, right now, as a society, invest in the relevant research so that we can understand what's actually going on inside these systems at the pace that we're developing them. But here's my question.

54:25Let's say President Xi Jinping comes to see Trump. And let's say President Trump was a different person and had a different view and decides he's going to make a pact with the Chinese. And we're going to create guardrails and oversee all of these things. A, would you even believe it? Right. And would it matter? We would need some mechanism for mutual verification. Similar things were true in the nuclear age. It's we we we can't take President Xi's word for, you know, they're really going to slow down and be as careful as possible. And so the devil's in the details of how you verify that this other party is doing what you expect to be doing.

55:01Ultimately, I think the key question is. But isn't what we're learning that even the ability to verify can be masked by the agents themselves? Meaning my understanding of when we were looking at some of the hugging face scenarios, some of these others, that that some of these agents were literally making moves, if you will, to mask what they were actually doing. And so the ability to verify could get more complicated. I ask this only because oftentimes people say, well, you just rip the cord out of the wall. And I think to myself, well, yeah, if you know that you need to rip the cord out of the wall, but you may not know to rip the cord out until it's too late.

55:43Yeah. So I think there are two important but distinct things happening here. The first is that when you're talking about verifying whether or not China is building a recursively self-improving intelligence, this takes a huge amount of compute, huge amount of power. And it is relatively easier to attempt to monitor that sort of thing than the case that you're describing, which is, let's say, individual AI agents, regardless of where they're deployed, understanding why they are behaving the way they do, why they may say one thing and end up behaving in another way. This is, I think, back to the point of the simple fact is we are building out incredibly powerful systems that even the people building do not understand how they work internally.

56:21They are like brains in that way that you have people behaving in very complex ways. And we don't understand the exact mechanism for why that happens. And that's why we need a neuroscience. And that's why we need an AI neuroscience. Our friend from California that really remembers we had down the Obernolte. I'm sorry. The guy Obernolte from California that is trained in computer science. We had him on. So he's still of the belief. He's Republican. But you still have the belief that, for example, a nuclear bomb, we weren't worried it was going to explode itself on us. OK, so it was the the players and the actors, the humans that were going to end the world through that.

57:03In this case, is it is it really the concern that the A.I. itself could could be the bad actor? And most people like his contention is that's not the way it's going to happen. That's not what we need to worry about. It's these tools in the hands of someone that want other humans, another country that wants to do it. We're back to that. Is it going to be someone else that's ahead of us in AI that is that we should worry about? Or is it the AI itself we should worry about? You're arguing it's the AI itself. I mean, I definitely think it's both. And we don't have to choose between being concerned about both of those things.

57:37We kind of do have to choose because, you know, we want to we want to stay ahead. I still think in the United States, we think we are morally on higher ground than China, just from human rights and Uyghurs. We wear the white hats. And we'd like to be the ones, you know, maybe with a different administration, you'd feel more. You'd feel like your hat was even, some people would feel that their hat was even, that we were on even higher ground. A lot of people think Trump versus China. They might pick China. But I still think that the United States is a force for good. And I agree with you. At the end of the day, I think what we need are responsible, wise human beings building responsible and wise systems that have the capacity to do open-ended.

58:23You're using words that don't describe a lot of leaders. Can I just ask you this idea about wise? When you look and see what a Dario says, a Sam says, or an Elon says on one end, and then you look at what a Jensen says, the president says, a David Sachs says, or Mark Zuckerberg says, who do you genuinely side with and what do you think is motivating them? I'm more sympathetic to the concerns of the lab leaders and I am to the people who I think stand to gain enormously from just letting this thing right. But the question is, doesn't, I would think Dario, Sam, and Elon would also stand to profit from all this.

59:05They're closer to the source of the action. What's going on at these labs is roughly two generations more advanced than what we're seeing out here. Sam Altman is talking about delaying OpenAI's IPO until next year. This is not the kind of thing that you would expect if what they're doing is some sort of massive regulatory capture psyop type thing. I think the most straightforward explanation about what's going on is they understand how powerful this technology is. They understand what the scaling laws are. And they are terrified about it because we are building systems whose internals we do not understand.

59:33I don't think the answer is to shut it all down. I think the answer is we need to scale our scientific understanding of what's going on in these systems before we just continue to let this rip. But who should set the rules? I think we need to see collaboration both between the labs who have the most expertise on what's going on internally and our government. We need intelligent, non-hysterical regulation to set in. So we do this when we deploy aircrafts or when we deploy drugs. Do you think the Bernie Sanders and others have basically said, just shut it down? Is that the hysterical regulation you're talking about?

1:00:08I think in the absence of understanding what's going on or having any confidence that this isn't going to go terribly wrong, slowing down this technology does make sense. The question is, slow it down for what? And I'm trying to provide a pragmatic answer, which is we need to accelerate our scientific understanding of what's going on inside these systems. They are not ordinary software. And if we do not build out a sort of neuroscience of these artificial brains, then we're going to continue to get surprised when they continue to misbehave in ways that nobody predicted. The other arguments we're having politically are starting to seep in, too.

1:00:42We had someone yesterday say that we need an EO executive order to prevent IPOs. Because once the company does the IPO, then it becomes a profit monger trying to satisfy shareholders and not for the. So you're seeing the whole socialist mentality even creep into arguments about about what we do here at this point. So they need to be funded, don't they? I mean, we need to allow IPOs to happen if if we want to continue a pace. Right. I think I think. The economic details of how these companies operate is less important than the core fact that the technology they're putting out is not transparent even to them.

1:01:27There's no other industry for which that is true. Do you not think it's really unbelievable that we're living in an age to actually see this happening? In the age of humans on this planet, we're at a point where we really are approaching the singularity where we have no idea, where all of human knowledge could pale in comparison to what we're finding out. It's unbelievable. It's unbelievable. It's science fiction, except it's not fiction at this point. And I think humanity has been telling ourselves stories about these sorts of moments happening for a while. Did any science fiction writers write really great things happening from this, or is it always bad?

1:02:04It's a bit of both. And I think we are on that tightrope right now. And I do think the hysteria and the panic I don't ever think is a pragmatic way to go about doing this. But the stakes are unbelievably high for not screwing this up. And it's going to take the wisest and most intelligent people making the best decisions we possibly can in the short term to make this not go horribly wrong. Great, because none of them run for office. So it's awesome. All right. Thanks. Thank you. Thanks for having me. And that's Squawk Pod for today. Thanks for listening. You made it. This has been a supersized Squawk Pod.

1:02:38Let us know what you think about our podcast or the topics we cover. You can write a brief review on Apple Podcasts or find us on X at Squawk CNBC. Drop us a line there. Squawk Box is hosted by Joe Kernan, Becky Quick and Andrew Ross Sorkin. Tune in weekday mornings on CNBC at 6 Eastern. Listen to Squawk Pod anytime you want. And we'll meet you right back here tomorrow. We are clear. Thanks, guys.

1:03:11This episode is brought to you by Schwab Market Update, an original podcast from Charles Schwab. Join host Keith Lansford for this information-packed daily market preview delivered in 10 minutes or less, including projected stock updates, monetary policy decisions, and key results and statistics that may impact your trading. Download the latest episode and subscribe at schwab.com slash marketupdatepodcast or find Schwab Market Update wherever you get your podcasts.