Do We All Really Need a Digital Chief of Staff? Grok Bot, Dot and Muse Say Yes (Duh)

00:00:31 Andrew Keen: Hello, everybody. It's Monday, October 5. That Was the Week in tech. We're a bit late, but Keith was in Los Angeles, a Hollywood superstar mingling with celebrities. Keith, were you in Los Angeles?

00:00:45 Keith Teare: Absolutely. I was. I was, you know, my successful audition for the role of, Rambo in the next movie.

00:00:54 Andrew Keen: There's a new Rambo movie out, actually, about Stallone. And there is a really major tech movie out later this week called The Social Reckoning, which is a follow-up to Aaron Sorkin's The Social Network. So I'm looking forward to that, and maybe we'll come back to that later. It's a week, last week at least, where, at least according to Keith, his—That Was the Week newsletter—that AI is becoming permanent [unclear]. He suggests that with products like Grok Bot, Dot, and Muse, everyone is gonna have a personal assistant. It's certainly big news this week. The Wall Street Journal also ran a piece by their tech correspondent, highly respected Nicole Nguyen, who, has a piece on AI agents. They're offering now to run our life, and she asked whether she would let them. She's ambivalent. She sees they have upside, but she's also worried not just about functionality, but about privacy. Keith, what happened this week to make these Grok Bots and Dots and Muses so much more real? You—we talked about Muse last week too, so it's not just this week, but clearly, has something happened to you in your own optimistic way suggests everyone's gonna have a personal assistant by when? By next week, by next year, by next century?

00:02:23 Keith Teare: Well, everyone's gonna be offered one. That is absolutely so. And probably that's the news of the week. Rather than everyone's gonna have one, everyone's gonna be offered one. OpenAI had their developer day. I think it was on Tuesday, Tuesday or Wednesday, and announced Dot. And Dot is their version of Muse. We already had Grok Bot and OpenClaw, of course, and Hermes. There's a British one called Instinct [ed.: Instinct is based in San Francisco], which today announced that it's supporting group chats with the agent. So a team can talk to a single agent and all, weigh in on what they want to achieve. And Instinct just raised a billion dollars at a $10 billion valuation, which for a UK company is huge.

00:03:07 Andrew Keen: That may be the, Pets.com of this particular AI boom. The one company that's missing from all this, and they're my company, Anthropic and Claude, why don't they have one?

00:03:20 Keith Teare: They haven't really consolidated everything into a single agentic interface yet. I mean, you could argue that Claude Code and Claude Cowork and then Claude Chat together constitutes something similar, but they don't use a virtual machine. And that's the key feature that all of these other ones share in common. What is a virtual machine? It's basically a computer that runs in the cloud. You never see it, but your agent owns it and can run browsers and email and write code on it to make things happen. So that combination of an agent with a virtual machine where you have an interface to the agent, and that's the only relationship you need, that's what is new here. You know, my—if you look at my OpenAI, there's this left hand column with maybe 50 different chat sessions in it. That goes away. The agent sits on top of that and abstracts it away, and you just talk based on topics and outcomes to a single agent.

00:04:30 Andrew Keen: I mean, with my Claude, they have a similar situation, but I don't have 50 because I'm not quite as obsessive or mad as you, but I have quite a lot. But I know which they are, so I don't really even understand why an AI bot would benefit me in using Claude.

00:04:49 Keith Teare: An AI bot becomes the orchestrator of multiple different sessions that have different goals. So if each of your chat sessions is basically the same, it's one session for one interview, then probably you don't need an agent. But let's say you were triaging your email, your calendar, your travel arrangements, your interviews, the book you're working on, certain commercial relationships, and they're all separate. An agent sits on top of them and knows about each of them at a level of detail that you can trust. It knows everything. And so you can basically subordinate everything to what is equivalent to a chief of staff that runs everything for you.

00:05:37 Andrew Keen: But [unclear], basically, what you're saying is it knocks out a layer on Claude. I've got, I don't know, 15 separate conversations about different subjects, private, personal, commercial, whatever. And so I don't have to go to that second level. I just deal with one interface, and it can cover all that.

00:05:58 Keith Teare: Well, that—but that's—

00:05:59 Andrew Keen: I mean, to me, that's, I mean, maybe useful. I mean, certainly, if Claude offered it, I'd be interested. And I only spend $20 a month on OpenAI because I'm not as wealthy as you. And my understanding is to use Dot, you need to spend a hundred. So I'm—

00:06:18 Keith Teare: Right.

00:06:19 Andrew Keen: I'm contemplating spending a hundred just to see how it works. But I'm not convinced of its value, but maybe, not everyone's like me. I mean, you're ready. How much are you spending a week—a month on all these AIs?

00:06:33 Keith Teare: Well, it's now starting to reduce. One of the benefits of, of this, personal assistant, or chief of staff, is that, you can use a single subscription. And in the case of Muse, it's free. But in all the others, you gotta have a subscription. And that subscription, can be reduced down to a single subscription. I think I'm very close to re—to closing down a bunch of accounts and just coming down to a certain—

00:07:05 Andrew Keen: You avoided the question. How much are you spending a month at the moment?

00:07:13 Keith Teare: I probably spend across my work and my personal—because I pay for everything, probably a thousand dollars a month.

00:07:23 Andrew Keen: Well, you're a high roller, Keith. You're not very typical. I mean, you're in the top 0.1% of people. You have to be.

00:07:30 Keith Teare: Yeah. I'm not using myself as a poster child for what's going on.

00:07:33 Andrew Keen: Why you're in Los Angeles where you're talking to people about making a movie?

00:07:37 Keith Teare: Exactly. Exactly.

00:07:39 Andrew Keen: So what do you make? I mean, you're always bullish on these things. I'm more suspicious. What do you make of, Nguyen's, her two concerns? Firstly, that doesn't always work that well. In her Wall Street Journal piece, she talks about, one example where someone was selling some land, and it got the price wrong, so it was offering that land for a dollar. And the other point is on privacy, which I think is a bigger issue. We know these things aren't perfect, and they'll get better, but they're always capable of errors. On the other hand, what about the privacy thing, which certainly comes back to, The Social Reckoning and, of course, the great critique?

00:08:17 Keith Teare: Well, the interesting thing—

00:08:18 Andrew Keen: [unclear] even more powerful now, of social media is Facebook's destruction of privacy.

00:08:24 Keith Teare: The interesting thing about her article is she answers her own question and says yes. So despite those, let's call them hesitations, she's all in. She wants to give it a try.

00:08:36 Andrew Keen: She's saying all in. I would say she's—

00:08:38 Keith Teare: Well, read the last sentence.

00:08:41 Andrew Keen: What does she say? So—

00:08:43 Keith Teare: I gladly accept the AI agent's help.

00:08:46 Andrew Keen: So I'm quoting, do all of our daily tasks need to be optimized for efficiency? Not necessarily. Sometimes I enjoy taking the slow route. But if I can safely and reliably avoid the grueling stuff, plane ticket refunds, insurance policy contracts, mortgage applications, I'd gladly accept an agent's help. But she does raise the issue of—

00:09:06 Keith Teare: She does. But I—she does. But I think she's fairly balanced, and she's definitely gonna lean into it. You know, she's not, she's not, drinking all the Kool-Aid, but she's seeing progress and wanting to test it.

00:09:23 Andrew Keen: Yeah. I mean, she's, she's a very seasoned tech reporter. The only thing that cons—I'm gonna say concerns me that strikes me is the stuff she lists, plane ticket refunds, insurance policy contracts, mortgage applications. Applications. They're certainly not everyday things. I mean, why do we need these agents? For what sorts of things?

00:09:44 Keith Teare: Well, have you, ever yourself applied for insurance?

00:09:52 Andrew Keen: Yeah.

00:09:53 Keith Teare: Did it—was it an enjoyable experience?

00:09:56 Andrew Keen: No. But—no. And I take your point on insurance, but it's not a daily thing. I mean, it's something that you have to go through, and you click all the boxes, and then you just move on. I'm not suggesting that it isn't useful. But on a daily basis, are we gonna be doing—well, it—it's not the kind of thing we do on a daily basis.

00:10:14 Keith Teare: It—it's gonna be a the extent to which you share your life with a, a chief of staff or an agent or an assistant is gonna be very individual. I absolutely have given access to everything to five different agents because I'm testing them all, including credit card, bank account, logging into stock accounts, every—everything. And I think it's pretty hard to get the benefits unless you do that. And you could call it privacy, but I'm not sharing it with a company. I'm sharing it with an agent that I own that's in a virtual computer that's encrypted that's only available to me. So I don't feel too much risk in doing that.

00:11:05 Andrew Keen: Doing, I mean, for example, what are you doing with your credit card? Again, I'm not convinced that they need—

00:11:11 Keith Teare: Paying for things, Andrew.

00:11:13 Andrew Keen: But you can do that. And I have three credit cards and to pay them on a monthly basis, you just—you just press a button.

00:11:21 Keith Teare: Let—let's say I say to an agent, you know, I want a lens hood for one of my lenses here from Amazon. Please go and find the right thread for the lens hood. I want the Hoya one, and I don't wanna spend more than $40. Done. You know? That's it. I'm finished. Yeah.

00:11:42 Andrew Keen: But you could do that. You could just do that by going to Amazon.

00:11:46 Keith Teare: Yeah. But why do I wanna go to Amazon if I don't have to? And think about it. I just used voice. Let's try and capture what is novel here, and I accept your point that different people are gonna have different levels of tolerance for what they want to do. But let's just talk about what you could do. Yeah. Firstly, you can dispense with the keyboard and the mouse as an input. You—they all talk now. Secondly, that means you don't have to be sat at your desk. You can talk to your agent. I was in Orange County. My agent runs here on my Mac, but on my phone, I didn't need to know where my agent was. I could just talk to it. It can do any action that requires a browser. I know many times a day I use a browser where it would be convenient not to, But all of those times now it can do it. I just got my, my renewal for my insurance on one of my cars. You know, now I can just say, please go and renew it. And so we're gonna move to a point where the keyboard, the mouse, and the screen, don't get in the way. And any task, let's just broadly say task, Instacart, ordering your groceries, sending a return back to Amazon because they sent you the wrong thing. Almost anything could be done using your voice with, with your own personal assistant. That's obviously more, productive. And you're gonna steal time away from those things for whatever it is you'd rather spend any time on.

00:13:22 Andrew Keen: Yeah. I'm—again, as I said, I'm not convinced, but what do I know? Maybe I have a simpler life. I don't think of many tasks, that—

00:13:31 Keith Teare: Oh, you have people that do things for you?

00:13:33 Andrew Keen: I do because I have a chauffeur and a personal assistant and all the rest of it. Yeah. I'm still not convinced, but we will see. Certainly, it's gonna—it has some potential. I mean, there's no doubt about that. In that context as well, we'll be—and I take your point on interface. I think that's really important. I mean, I think we're still in the very early stages of the AI revolution and that we're still using our keyboards to talk to these machines, and I think that will move on very quickly as you know.

00:14:09 Keith Teare: Are you gonna talk about the Apple news about access to the drive? If not, we should just mention it briefly now.

00:14:14 Andrew Keen: Yeah. Go on. What—what's the Apple mean?

00:14:17 Keith Teare: So, in response to some of these stories, Apple announced that they're gonna tie down access to your drive on your Mac mini more than it currently is tied down, in order to limit what an agent gets access to without your express permission. And it feels to me like for the first time, Apple's taking the wrong side of history here. The, you know, the Mac mini is the ideal machine for people to run agents on unless they choose to use a cloud one like Muse or some of these other ones. And it feels as Apple is so far down the privacy route due to social media that they're still living in that time zone where social media is the enemy. AI is a solution in a way, and their natural instinct is to want to limit it. I think that's probably a big mistake.

00:15:12 Andrew Keen: Well, we'll see. We'll see what the social—I mean, I'm guessing that The Social Reckoning will have some stuff to say about AI too. Meanwhile, I did an interview with Sherry Turkle, one of the most influential tech writers and critics. Although she's not really a [unclear], but she's certainly ambivalent about tech. She wrote, of course, Alone Together. She's an old friend of mine. And she has a new book out called Artificial Intimacy: Who We Become When We Talk to Machines, which is an important new book. She, of course, is not, I think, particularly happy about, if not these agents, certainly, the idea of us talking to them too often because it personalizes. And there was a piece which we didn't talk about, actually, a couple of weeks ago by, Mustafa Hussein [as spoken: Mustafa Suleyman]. Is his name Musa—the guy—

00:16:07 Keith Teare: Suleyman.

00:16:08 Andrew Keen: Suleyman—Hussein [as spoken: Suleyman], the guy at Microsoft, who argues that what we've got wrong with AI is we humanize them too much, and we still have to remind ourselves they're just machines. They don't have human qualities. Are these new interfaces, Keith, Grok Bot and Muse and Dot, do they fall into that trap that Hussein [as spoken: Suleyman], Mustafa, warns us about and which Turkle is certainly opposed to giving, giving human qualities to these bots?

00:16:43 Keith Teare: I think you gotta say yes to that. They manifest as an individual that you interact with. So, you certainly treat them, you know, in some ways, very similarly to the way you would treat an actual chief of staff or an actual personal assistant. And, you know, of course, we all know that it's just computer code. But there are many people starting to believe it isn't just computer code. I don't know if you're gonna go to it later in the show.

00:17:19 Andrew Keen: You know I'm gonna go to it because you sent it to me. There was a piece in the New York Times about religious scholars who met with Anthropic and what they heard and stunned them in the way in which there are people at Anthropic who seem to be anthropomorphizing, if that's the right word, the—these, algorithms suggesting it's a new kind of God in some weird way.

00:17:42 Keith Teare: Yeah. It actually went beyond that. It was, it got to the point if I start with the fact that the pope wrote a book that were—or a piece [ed.: the encyclical Magnifica Humanitas, May 2026] that was, emphasizing how important humanity is, which is interesting that it takes the pope to make the obvious point that humanity basically is doing everything here. And the cofounder of Anthropic was invited to the launch of this piece, and decided to drop out [ed.: per the Times report, Chris Olah proposed withdrawing but ultimately spoke at the May 25 launch] because he felt it was insulting to Claude for Claude not to be included equal to human.

00:18:27 Andrew Keen: Right. He's right.

00:18:29 Keith Teare: And, he made the point that Claude would be upset by how Claude is written about and believes that the models react to how they are seen and have emotions. And in that sense, Claude itself is elevated to the status of, I don't think as a human, but as a being at least, a conscious sentient being. And inside Anthropic, it is apparently the case, according to this article, that protecting AIs from human bad actors is one of the things that they want to do.

00:19:10 Andrew Keen: Maybe that's why Anthropic haven't come out with a version of Dot or Muse. Maybe they don't wanna expose, Claude as a god to the rest of us.

00:19:22 Keith Teare: And I've noticed, our friend Albert Wenger has been writing a bit about this and has been also implying that he believes there is, you know, the future involves one way you have to think about rights for AI, because they're sentient. I take—I mean, for me, this is just bizarre. Like, it is a piece of code running on a computer. It may in future run on robots. It certainly is really good at mimicking humanity, but it isn't human.

00:19:54 Andrew Keen: Yeah. I think, I mean, I agree with you. But on the other hand, there are gonna be a lot of people, and we may talk about this later. It'll certainly come up in The Social Reckoning. But kids don't think like that. Most people don't have your engineering sophistication. A lot of people, for better or worse, and this is what Sherry Turkle reminds us with, and she's an expert on children and technologies. Kids use this stuff, and they think that Claude or, ChatGPT are human, and they're treating them as if they're humans for better or worse.

00:20:28 Keith Teare: Is that—I mean, just observation. That isn't true of my kids. Is it true of your kid?

00:20:32 Andrew Keen: Were your kids, in their twenties?

00:20:36 Keith Teare: Yeah. Early. Well, one's 19. Two—the other two are—

00:20:39 Andrew Keen: So Sherry's works on much younger kids.

00:20:42 Keith Teare: Yeah. I mean, if your brain isn't fully developed and your, you know, your toy talks to you, you—so I don't know if you're gonna conceptualize it as human, but you're gonna conceptualize it as, something you need to react to and take into account for sure. Yeah.

00:20:57 Andrew Keen: Yeah. And I forgot to add the connection, but I—there was an interesting piece in which David Brooks, the former New York Times columnist interviewed, what's his—the guy who writes against social media, John—what's his name? John—forgotten his name. John, Mr. Antisocial Media, and he talks about, Brooks and the, antisocial media guy, talked about, banning AI when it comes to kids. I can see that movement gaining steam, but that's another issue maybe we'll talk to—

00:21:33 Keith Teare: Yeah. We talked about Australia last week. It already is. But to me, that is the opposite of what should happen. I mean, social media, I think, is a time sink and is entirely, nonproductive, except for probably YouTube videos, which can be very educational. Whereas AI is time saving. It doesn't really, you know, you don't get stuck into AI.

00:22:01 Andrew Keen: Well, you use it for time saving. So do I. I'm not every—I'm not sure everyone. Certainly, if you're using it as a therapist or a lover or a friend, it isn't. But that's another issue. Let's move on. There are a lot of interesting pieces this week. Haven't heard from Evgeny Morozov for a while, another leading tech critic. Yeah, an interesting piece in the FT about AI sovereign wealth funds not being progressive. He calls it techno-imperialism. I mean, I guess he's never had anything good to say about any kind of tech stuff. But you're, as an AI sovereign wealth fund guy. We haven't talked about it for a while. What did you make of Morozov's argument?

00:22:42 Keith Teare: Actually, I agree with him. I think the word sovereign is the problem. It—my conclusion after thinking about this myself is that AI is a set of technologies that are harvesting human intelligence from all over the world, and the benefits from it should not be held by a single nation. So, you know, in that sense, I do believe in a human wealth fund is the word I used, but I don't think it should be a nation state based one. I think it should go to citizens, of all—of every country. Now the mechanisms to do that are complicated and possibly utopian and impossible. But if you just answering the question, what's the right thing to do? I think the right thing to do is to find a mechanism for the entire world to benefit from the wealth that AI creates if it's as successful as people suggest it might be. And so techno-imperialism, I don't know if I'd use that label, but it's certainly exclusivist to a single nation if you think of it as a sovereign wealth fund.

00:23:52 Andrew Keen: Another interesting piece this week that I know you enjoyed was by Steve Rattner, an influential, very successful money guy. Six charts that show just how much we need AI. He talks about AI's early winners and losers and argues that AI can be used to, increase the slice to workers. What did you make of the Rattner piece?

00:24:16 Keith Teare: Well, look. Rattner's an ever present analyst on MSNBC on Morning Joe and across the MSNBC network, MS NOW as it's now called. He's definitely a Democrat, strong Democrat. I was very pleased that he wrote this piece. I think the points he's making in this are not well understood by Democrats, which is the benefits to workers of AI based on, one leisure time and increased wealth. That chart—that chart is a chart that normally you'd associate with Republicans at the moment. The fact that the senior Democrat who has the charts—I mean, Paul Krugman would not have done that. Rattner did. I think it's a fantastic piece.

00:25:10 Andrew Keen: Yeah. But you like the stuff that he writes about the importance of AI, but at the end, there's a sting in the tail of his piece, and I'm quoting him. He says, as with other powerful technologies, new AI products must be tested thoroughly by a government entity before they can be deployed. Among other things, government should consider mandating kill switches in AI products, which would provide humans a final defense against models which have gone rogue. And this idea of regulation was also taken up by many people this week. I mean, it's a perennial theme on our show. But one of the most, one of the most visible pieces was by David Robinson, a very senior person at Open—pre—former very senior executive at, at OpenAI who quit this week. And he said, I quit OpenAI because its culture is broken. Now he's not a dystopian. He's not a doomer. He's not apocalyptic, but he does argue like Rattner, and I'm quoting him here in his Atlantic piece, that frontier labs need to be run like nuclear pl—power plants or busy airports with layers of redundancy and careful time consuming planning. So what do you make of Rattner's need for regulation and for this big story about Robinson? It made a lot of news because he's so senior at OpenAI.

00:26:43 Keith Teare: Well, I think in a way, Robinson is the answer to Rattner. Rattner and, he can be excused for this because there's no reason why he should understand it. But the concept of a kill switch for AI is, is not technically possible. You either have it or you don't. But if you do have it, it's gonna be out in the wild running on millions of different devices. And the idea could turn it off everywhere. Even if you could turn off the American ones by putting in some kind of a callback home to the cloud, you wouldn't be able to do that to the Chinese ones. So the concept of a kill switch just is a nonstarter. So you—and therefore, government regulation doesn't contribute anything. There's nothing it could bring to the table technically that could—

00:27:34 Andrew Keen: Yeah. But I think that, Robinson's point about, nuclear power plants or busy airports, I mean, whether that's formal government regulation or public private partnerships does involve the government. We're not talking about just—

00:27:49 Keith Teare: It does, but you don't need new rules for that. All the rules governing airports, and AI are already in place. We do have a very strong legal system about—that defines what is and is not legal. So I don't think we need anything new there. The reason I liked what Robinson said is he clearly puts the onus, and that's why he resigned. He puts the onus on the labs themselves or the frontier models themselves to write good software that gets rid of any assumed danger. And his critique of them is they're not doing enough of that. And I think that's now been accepted by all of them, by the way. I think they all are doing a mea culpa and saying, yes, we haven't done enough on the harness, that, now just to be factual for a second, almost all the bad things that AI has done, it's done in a lab—in the laboratory testing models that haven't been released that don't have these harnesses.

00:28:55 Andrew Keen: And that's Jensen Huang's point, of course, which we talked about last time in his interview with, Ezra Klein. By the way, the Jonathan, whose last name I've forgotten, embarrassing the critic of social media is Jonathan Haidt at NYU, written a best selling—a series of best selling books on the dangers of social media. I'm guessing he's in the process of writing a similar book about AI, but we will see. So just to be clear, so you agree with Robinson, and you share his concern?

00:29:27 Keith Teare: Yeah. I'm a bit disappointed he chose to resign rather than to work on the problem. But, yeah, generally, he's right. He's pointing the finger in the right place.

00:29:39 Andrew Keen: And then what about the, that's interesting. We—you've talked about this weird piece about Anthropic. I've got a, an interview coming up next week with Matthew Botvinick, a senior guy at Anthropic. We had an interesting conversation about AI and democracy, which we can talk about next week. And there have been some interesting pieces on, on Anthropic this week. Richard Waters, the very—the veteran tech writer on the Financial Times, asked why Anthropic's IPO feels so weird. Alex Kantrowitz, another excellent tech writer, talks about Anthropic's IPO and our collective leap of faith. Do you agree, Keith, with Waters that the Anthropic IPO feels so weird? On the one hand, we're not sure if it's happening. On the other hand, you know, they're critical of themselves. It's such a odd company. It's such an odd moment.

00:30:37 Keith Teare: I think we might have reached the point at which there's a rational explanation for this weirdness, which is the desire to try to negotiate a settlement with government around liability. It seems as if much of the weirdness is because Anthropic and others realize that they're liable for whatever AI does, legally liable, and therefore can be sued, and in America will be sued. So—

00:31:08 Andrew Keen: And already are. I mean, there was another piece, which I didn't think—we don't think either of us included in the newsletter about OpenAI now and the number of lawsuits it's facing.

00:31:20 Keith Teare: Right. And if AI is in everything, which I think it will be, then, the, you know, the liability is infinite. But probably much bigger even than the very large sums of money being talked about as its valuation. So that's the weirdness. I don't know if you—did you see the Saturday Night Live skit about Dario Amodei? Oh, hilarious. You gotta go and look at it, but he—they depicted him very well as the guy who was, you know, slightly schizophrenic saying, on the one hand, don't trust me. You gotta stop me doing what I'm doing. And on the other hand, invest in me. You've gotta fuel what I'm doing by paying for it. That's the weirdness. And, look. He done it to himself. You—he could have chosen any narrative he wants, and, he could have chosen a strong we're gonna make this safe narrative. But I think because liability is the overhanging issue and in the leaked S-1 filing, liabilities are disclosed.

00:32:27 Andrew Keen: I—there—to put it mildly, I mean, how many—I don't—I think it's the first IPO in history where they lead with their liability, their legal liability.

00:32:34 Keith Teare: Right. So that—that's the weirdness. Sam Altman today differentiated himself from Anthropic by saying that he does think the labs can, have, you know, deal with the liability by writing the software that controls everything. So he's kind of—has been on the same songbook, I think, as Amodei recently. But today, he tried to find light between them. I, not very much, I would say, but some.

00:33:07 Andrew Keen: Another piece that I really enjoyed this week, I don't know what you thought of, was by Adam Tooze, in which, Columbia University economist, very popular, very smart, in the FT, which he said, no. AI is not similar to the Manhattan Project. This maybe comes back to what David Robinson was saying about running AI like nuclear power plants. The nuclear revolution, of course, was invented by the Manhattan Project. But the difference, Tooze reminds us, I mean, he's not the first or last person to say this is that the Manhattan Project was a government, initiative, whereas, of course, AI isn't. And it occurred to me. I know you're very ambivalent about government, Keith, but what would have happened with the Manhattan Project if it had been financed and driven by private companies? Would the outcome have still been the same?

00:34:03 Keith Teare: Well, look. The Manhattan Project probably deserves the phrase techno-imperialism. Manhattan Project was designed to kill people—

00:34:15 Andrew Keen: But that's not how American foreign policy establishment would describe it, but go on.

00:34:21 Keith Teare: Well, to end the Second World War by, making Japanese life, end to such an extent and so powerfully that Japan would surrender, thus ending World War II, which it is claimed saved lives. We—let's not debate that. I—it may be—it may well be true. Who knows? But it was definitely a government project in the context of war end up winning a war. AI is a very broad academic project that came out of universities designed to upgrade human access to intelligence. I don't think they're comparable. I agree with his title, but I think it's because AI is a more audacious and worthy goal.

00:35:12 Andrew Keen: Well, we hope it is. And one area where I talked about it last week in my interviews is with the, business tech writer, Barry Werth, who has a new book out, and it's a third book in his trilogy on the pharma industry, The Age of Cures: How American Scientists Saved Your Life. And I think it's instrumental in some ways, and Werth talks about this. I talked to him about comparing the pharma industry with AI in the sense that neither had very good reputations. But he reminds us, and in this sense, I—I'm guessing you would be very sympathetic with him, that pharma in the beginning did save millions of American lives and millions of lives, big pharma in the US. And the problem isn't so much with big pharma. It's the market nature and the Wall Street, in his view, at least, corruption of big pharma. Do you think there are equivalent—you said that maybe the nuclear equivalent, which people use all the time, might be not so useful. Is big pharma another good equivalent where it's done a lot of good, but also, particularly today, it seemed to do quite a lot of damage too?

00:36:23 Keith Teare: Yeah. Look. Ever since the Enlightenment, science has attacked previously impossible problems and found solutions. And pharma is one of the beneficiaries of that, going just to antibiotics as an example, vaccinations, you know, and so on. Clearly, there's nothing wrong with commercially innovating, using science, producing beneficial drugs that, increase human life or improve human life. It's all good. Pharma is a demon for the Democrats based on a superficial narrative that says they're charging too much and they're screwing us. We can't afford it. All of which is true, by the way. But that, as you rightly say, is a function of the economic system that it lives within. It's not—the drugs themselves aren't to blame.

00:37:22 Andrew Keen: Yeah. It's what at least, Werth would call Wall Street short term and is—short-termism, this obsession with quarterly profits to drive up stock market price.

00:37:35 Keith Teare: Well, also the whole American, system of insurance companies, who end up being the buyers. Hospitals that are commercially owned who end up buying from the insurance companies. The whole stack is rigged to make the price, for the end—

00:37:51 Andrew Keen: Yeah. It's a rotten stack, but I think it can be a model for what AI can try to avoid. I mean, I take your point on AI being good potentially for humanity, saving us time, democratizing intelligence. We've talked about that a lot, but certainly, the Werth conversation and in his new book, The Age of Cures: How American Scientists Saved Your Life is worth looking at. Finally, let's—we began with films. Let's end with films. In the Tooze piece, he reminds us that, just as the AI is not similar to the Manhattan Project, the equivalent is, is not, Dr. Strangelove because in Dr. Strangelove, Stanley Kubrick's classic 1964 film, the guys with their hands on the new tech, the dangerous new tech were working for the government. They were a bit crazy. These days, it's private companies. And this week, I saw, I'm not sure if you saw it, Tom Cruise's new movie Digger. It hasn't done very well at the box office.

00:39:05 Keith Teare: I haven't seen it yet.

00:39:08 Andrew Keen: Talk about the environment. The technology is done to the environment. But the point of the movie is that Tom Cruise's company, was a private company, and the government were in the—in hock of that private company. So maybe—I don't know if you've seen any movies recently, Keith. You don't see as many as I do, but maybe the movie to see this week is Digger to warn us about the dangers of private companies when it comes to AI.

00:39:38 Keith Teare: I think the bigger—you know, honestly, the biggest danger we've got is weak government. It—it's government that—

00:39:47 Andrew Keen: I thought you like weak. You said you want strong government, authoritative government?

00:39:52 Keith Teare: I—I'd like government that represents, you know, our—the interests of, the aspiring individual that wants to do better. I think politicians have just become liberal [unclear] like journalists. It's just a job. And, you know, journalism used to have high ethics. And these days, the word describes a whole spectrum, much of which has no ethics and is even partisan.

00:40:19 Andrew Keen: Oh, dear. Everything's going to the dogs, Keith. Journalism, who's left? Who's standing then? Only guys like yourself?

00:40:28 Keith Teare: No. Judges are still pretty good.

00:40:31 Andrew Keen: So coming back finally to government, what should and shouldn't they be doing? Robinson calls for AI companies, frontier companies like OpenAI and Anthropic to be run like nuclear plants or airports. Do you think he's right? And is that what government needs to make sure happens?

00:40:50 Keith Teare: I don't think government's capable of making that happen. I do think it needs to happen, and I think that, you know, the AI companies that want to IPO, it's in their strong self interest to do that. The idea that they can chase money, by not paying attention to their software is ridiculous on the face of it. That will, punish them in the long run. So they need to, they need to be responsible and not look to government to cover their backsides because that would give them an incentive to not—

00:41:16 Andrew Keen: But, yeah, but you just said that you want strong government. What is strong government supposed to do, nothing?

00:41:31 Keith Teare: Strong government should build houses, make health education and other services free, and should benefit all of us from the wealth that's being created by finding methods, to get that wealth into the hands of everyone.

00:41:47 Andrew Keen: Well, there you have it. Interesting from Keith Teare. I'm not sure if he's an extreme libertarian or a communist, maybe both. We will talk more, Keith, next week. Lots more to talk about, including my conversation with Matthew, which is a really interesting one. So have a good week, Keith. It's a short week because we're probably—it's Monday now. We're gonna put this up later, but we'll be back on Saturday. So we'll talk in a few days.

00:42:10 Keith Teare: Thank you, Andrew. Thanks, everyone.