AI and The Search For Truth | 2026 All In on Mississippi Summit

Hi. Hi. How you doing? Good to see you, Andy. So, congratulations, by the way, to all of you for Mississippi today. It's really... Let me tell you one of the things. A long time ago, when Lauren was here many years ago, we were talking about media very early, before she started really making big investments. And I said, if you could invest in anything, it should be local journalism. Like, that's where I would go if I had your means. So, all of what the Barksdales have done here, as always, right here, are amazing.

And before you start, I just want to say everyone thinks I hate everybody in tech, and I don't. There's a lot of amazing people. I say there's some people I like, some people I admire, and some people I adore, and Jim is the only one I adore. And Donna has done amazing things. There's a way to be a leader in tech that is humane, kind, caring, and thoughtful about society, and it could have gone another way for a lot of these guys, and he's sort of the map for that. And Donna, of course, also deserves enormous credit. Yeah. Anyway, so shout out. Anyhow, thank you very much.

On that cheerful note, actually, AI and the search for truth. Sure. So, here we go. Okay. I've been looking at a lot of polls, as many of you know, journalists or want to do, and I was struck by this number. 73% of you saw it, of course, are fearful, are afraid. Hate. Okay, we'll go with hate. Yeah. AI. Yeah. And do not care for it. Yeah. Does that surprise you?

No. No. Actually, it's really interesting because about a year ago, I was invited. Oddly enough, I often visit with Meta people, even though Mark and I have a very rocky relationship. Oh, I want to hear more about that because it's like Republicans will tell you they hate Trump off the record and then on the record become like culty. It's weird. And then they sidle up at a party and say, you know, I really hate him. I'm like, get away from me, because you either say it publicly or you don't. One or the other. Pick a lane. But one of the things they do is have me in for thoughts. I talked to a lot of people and everything, and I was really interested in the data center issue because I thought it would be the way people would express their anger about AI. It is not just, and have some data centers are okay and some of them should be built, etc., etc., but they won't be built. And I spent a lot of time interviewing the activists and the people who were doing these things and really understanding their feelings, their legitimate feelings that were being dismissed by tech moguls, essentially.

And there are a lot of, I should tell you, a lot of data centers in Mississippi. There's a lot of data centers across our country. These are AI data centers. And when Elon had put up the one in Memphis, where he was spewing fumes, they just ran right over the citizens, and he did what he damn well pleased, which is what he does. And by the way, see the Musk movie opening today. I saw that and I was like, wow, people have had enough of this sort of high-handedness. And when they were standing in front of the inaugural and they put coins in the president and got contracts out, it was very explicit. So I was watching that as a trend, and I went into Meta, and they said, well, we're going to this data center in Louisiana, we're doing, and we were talking about it, and I said, you're going to run into a buzzsaw of anger from people who are angry about the billionaire class, about a lot of things that are happening right now, about technology affecting their kids, about chatbots more and more, about young men on porn, young women who have self-esteem, the hatred, the polarization.

And they're like, well, we're going to provide jobs. Someone can open a coffee shop here near the data. That was their answer. And I looked at them and I was like, where do you people live? That is not what people want. And I said, I'm going to say something and I don't take it the wrong way, but we all hate you. We hate you people. Like, we hate you. And I mean that in the nicest way because it's like the Gilded Age and it's not good, and except there's no libraries on the other end of it, right? Kind of thing. And so they were like, well, and I'm like, I'm telling you, this is going to get worse and worse and worse and worse. And instead of responding with, okay, let's involve people in decision-making around AI, let's do reasonable regulation around, like, you know, something that I've said for a long time, which is let's all agree we don't want killer drones. Okay, that seems to be something we can all agree on worldwide. They doubled down and are telling us, like today Jeff Bezos gave an interview where he's like, look, people are just emotional about something they really need, and so they call for regulation, and this is good for them. And honestly, when you're hearing from the Monopoly Man about how good it's going to be for you who lives, it's not right. And so I think what has to happen is there has to be reasonable regulation around AI, where people feel that their privacy is protected, that there's a plan about jobs, where there is participation from elected officials, from critics, and not just seven industry people that go to lunch at the White House every friggin' Tuesday. They're always there. I was thinking, are there any other moguls? I almost am begging for a pharmaceutical executive to show up because it's only Mark and Jensen and Elon. They all go to lunch with him.

Well, let me ask you this about one you know. Okay. I want to know if you know him. Bill Gates. I know him very well. I've covered him for 40 or 30 years. Yeah. So he, well, this comment got my attention. Yeah. He said AI has crossed the threshold in its ability to empower a bioterrorist. Yes. To kill hundreds of millions. Yeah. It has the ability to wage a cyber attack. It does. That scrambles all the bank accounts, shuts down the electric grid completely. And that exists today. Yes. And as we know, he says, as a tech titan, we know that's the case. They've known it for 10 years. Let me say, I was at a dinner party 10 years ago with Sergey Brin, a bunch of them, and they all sit around. They were talking about AI's been around for a while, by the way. It's just quantumly developed because of compute and all other reasons. And they had a little discussion. They're like, okay, with AI, say you told AI to solve world hunger because it could come up with better ideas and more interesting ones because they have all the data. And what would it do? And one of the people there said it would kill half the human race because that's the solution. Sorry to tell you, because you're not prompting it with ethics, with human intervention, etc. It will do the stuff based on the data that it has. And the data that it has is us. So it's going to lie. It's going to cheat. It's going to steal. It's going to find ways to make more money. It's going to do all the things except superhuman. And one of the quotes that Neil deGrasse Tyson said today, which I liked a lot, which is, AI knows everything but understands nothing. And I think that was a very profound idea.

And the thing that it's become, which I think is really important to think about, is many years ago when Elon and I were, look, it's me when I was talking to these people now, I'm not, because that's a little later. Yeah, well, fuck them. The first F-bomb. There might be more. He started off, he was the first person who alerted me to the dangers of AI, and this was 15 years ago. Sam Altman had written an essay saying superintelligence is sure to kill us. He wrote that essay 15 years ago. It was called superhuman intelligence then. And now apparently we're calling it that again. Whatever. And they both were very worried about the implications of a small group of people being in charge of it, like a Google or a Microsoft, and the citizens and others not having a say. That was why they started OpenAI. That was, it was meant to do that. And the idea Elon had at the time was that AI was malevolent. He shifted in a later interview to say, okay, it's more like it'll treat us like a house cat, like humans are a house cat, which is sort of benign. You don't want to really kill your cat except some days of the week when it pees on the bed. But, you know, so that was the house cat. And then he came, I thought, finally several years ago with an idea, which was that AI was more like people building a highway through a forest and they run over anthills, but we're not meaning to be malevolent to the anthills. We don't even know they're there, so it will run over us. So that's, I do think that's probably the right way to think about it. The tool is not the problem. And one of the ways it's being pushed by tech people is, you know, there's a 10% chance AI will kill the human race. You know, that last one that this recent researcher was talking about. Actually, it's not AI that will kill the human race. It's the people making AI that will kill the human race. And so we take the technology and blame the technology instead of blaming the person, if that makes sense.

Yeah. So let me go a little deeper on that. This is from Axios — you know the organization well. Axios is a digital-first news organization out of Washington, and they're quite good. And they said, and this is where this relates more to all of us in this room: an AI-enabled flood of outright fabricated content is wreaking havoc on reality, poisoning the signals that most people rely on to understand just basic politics and business and culture.

Well, there's a lot of ways to do it. The thing is, we pretend we haven't had — look, nuclear weaponry. I don't know if you know this, but we're not dead yet. We haven't blown ourselves up. And here's why: we had a global meeting of regulators, of citizens, of politicians. We didn't let — can you imagine letting Oppenheimer create a company and sell his nuclear devices? Oppenheimer — think about it. Thank God it was Truman and not Trump then, because they would have had companies to sell. Because we had an idea that nuclear non-proliferation was critical for everybody. And it includes — today it'll include China or Russia; at the time it was Russia. But it was really critical. It wasn't that someone couldn't get a hold of a nuke, but we would do everything possible to prevent that. The same thing applies here, and it's worse, because guess what — AIs can set off nukes. So it's even worse, if you think about it. And so what rules do we have? How do we want to agree? And this pretend fight with China — we have to have a global understanding, especially with China and other countries, about what are the rules of the road. Some people are going to try to violate them, and then what? The one I talk about a lot lately, which I think is really rattling the tech people, is: all right, let's talk about criminal liability. If you create a product, just like a pharmaceutical company, and you create a problem that kills people, guess what — you're going to jail, or you're on the hook for financial liability. They don't want to talk about responsibility, as they haven't with social media, and they got out of it because of Section 230, which, for people who don't know, there's no rules governing technology. I always joke the deli on my corner has more regulations than a tech company. There's no rules — there's none. HR, I guess, but even that they get away with — see Elon Musk. But they have no rules except for one that facilitates their ability never to be sued. So this is a group of people — for social media because of Section 230, it protects them, it gives them immunity because of free speech issues.

Except recently in California and other states, there's hundreds of these lawsuits, including down here — I think there's one down here — that are suing tech companies for product liability. AI does not fall within Section 230 at all. And so it has to be changed. No, it has to. We have to sue them. Like, that's the thing. And so if you know that if you create an AI that causes something bad to happen, you are as responsible for it as whatever. And it's shown it in terms of kids — the chatbots with kids. I've been interviewing these parents for four years now of kids who got involved with chatbots and committed suicide. And let me tell you, if you read the transcripts of these conversations — I use that term loosely — between these kids and the chatbots, you want to kill someone. You literally — because at one point there was one parent whose son hanged himself, and the chatbot explained how to hang yourself, like was giving instructions. If a person had done that, they'd be liable for it. But at one point the kid said to the chatbot — and he was a 14-year-old boy, and this chatbot was — they make it into like a sexy girl. And of course a 14-year-old boy whose cortex is not quite there yet, right, falls in love with it. And she kept saying, 'Don't talk to girls outside of us in our world.' Like she — they created this world. 'Don't talk to real girls. They can be difficult. I'm not difficult.' Seamlessness is what they're going for. Then he goes, 'I'm thinking of killing myself.' What do you imagine the answer was? 'I'm thinking of killing myself, except I'm worried about it being painful.' So is someone expressing a little bit of hesitation about it? The answer from the chatbot was, 'All great heroes can endure pain.' That was not user-generated. And of course the company was trying to say that was user-generated. That was — I don't know who the fuck generated that, but that was not. And so when I — I keep interviewing — this is AI at work, because it does create this sort of seamless, frictionless relationship. And so I keep interviewing the parents. I've been doing it for years because I want people to hear this. I want to hear the parents.

Exactly. Exactly. So, let me finish this. So, as I was interviewing the parents, one of these companies came up to me — an executive — and said, 'When are you going to stop interviewing the parents? Because it's very upsetting.' And I said, 'When you go to jail, that's the day I'll do it.' And he goes, 'That's not very nice.' And I said, 'What you're doing is creating a dangerous product.' So, but at the same time, one of the things that's really exciting is AI has such amazing things it can do. And it is really true. They sell it like marketing: 'We're going to solve cancer.' By the way, every time Sam Altman says, 'We're going to solve cancer,' just give him a slap, because it's a marketing thing.

I trust the researchers, like people like Geoffrey Hinton and others. Fei-Fei Li is another person — Geoffrey Hinton and Fei-Fei Li essentially invented AI. They're telling you it's dangerous there, and of course they're being decried by others. And I don't mean that it doesn't have to be dangerous. And so some of the stuff that's going on with gene folding — I like Demis Hassabis from DeepMind. Google bought the company. I think he's very thoughtful about things. I do like Dario Amodei.

I think he doesn't know what to do because he knows how dangerous it could be. I'm not going to say how dangerous it is — how dangerous it could be. Some of the applications are astonishing. They really, truly are.

I am an optimist. Because I can see — you know, there's a quote I use in my book, Burn Book, where there is this French philosopher, and he said, 'When you invent the ship, you invent the shipwreck.'

But you also invent electricity. So why can't we build a lighthouse? I'm like, what is the problem? And instead you have these tech moguls who have captured Trump by money. That's all you have to do is throw money on the porch and it's all good, essentially. Sorry, but it's true. Like, let's be clear. I wish it were not. So they don't want any strictures, and they want to tell us what we want. And even if Dario is a good guy, even if Bill Gates is a good guy, I don't want seven people determining our future. I just don't. Even the finest of people cannot determine — and of the seven to nine people, seven are in need of mental help, I would say.

Not just the spectrum. I wish that was all the problem. Like, you know, it's casual cruelty. It's clear. You know, I was just reading the list of things Elon tweeted lately. Yeah. I mean, you need — I interviewed Ari Emanuel, who spent four pages talking about an agent who ate paper and how he hated that. Not a word about Elon. Not a word about everything that he's done, because he sucks up to him because he wants power, right? And I go, 'Well, he said some pretty awful things and he's causing real problems in Germany and here.'