Shut Up or Shut Down? Amodei, Altman, Musk & the Great AI Freakout

00:00:31 Andrew Keen: Hello, everybody. It is Saturday, September 12, 2026. I'll probably put this out tomorrow on Sunday. Earlier this week, we did a show with a Finnish writer, on, zeitgeist, what he calls waves of change when everybody changes their mind and it inaugurates a new moment in history. And I wonder if we're living through that today. All of the big papers from the Financial Times to the Wall Street Journal to the New York Times are all leading with the same story that, Sam Altman and Elon Musk and Dario Amodei are all on the same page now calling for an AI slowdown. I think this initially came from a piece by Dario, and, Altman and Musk, who in general, particularly Altman, not particularly friendly or sympathetic to, Altman's, to, Dario, have jumped on board and said that, they need to slow it down. And this comes on the back of what The Wall Street Journal calls America's great AI freakouts, covering all sorts of things. The Journal says that Congress is suddenly waking up to the AI doomsday threat. Whether that's real or not, they're waking up to it. And, Axios had a scoop earlier this week that Mike Johnson, the Speaker of the House, is being urged to cancel the House recess over AI warning. So, Keith Teare, the author of the That Was the Week newsletter, has never been particularly friendly to slowing down AI development, which, we've talked about in the past. His newsletter this week is "AI Is Grown", using all sorts of biological metaphors. But I wonder, Keith, we know you're not in favor of regulation or slowing down. We've talked about that probably too many times. But, come on, do you think, assume that this time it's for real, that, the fact that all the main three players in AI are on the same page calling for a slowdown, does this represent a new chapter in the history of AI?

00:03:04 Keith Teare: No. I think it really it's really illustrates the weakness of AI leadership and the fact that they can be influenced by a six week long employee who resigned and wrote an essay and has been on TV and—

00:03:19 Andrew Keen: Yeah. But that's not my question whether you approve of it or not. As I said, you always, for one reason or another—

00:03:25 Keith Teare: You don't seem to be answering whether I approve. I was diagnosing the cause.

00:03:31 Andrew Keen: But in historical terms, you think in historical terms, whether or not you approve of it, whether or not it reflects the weakness of AI leaders, we know you're not always a big fan of them, particularly Amodei. Does it suggest that, and especially given the political tremors associated with this about Congress waking up, does it suggest that regulation is now really, for better or worse, I know you're against it, on the horizon?

00:04:02 Keith Teare: My editorial this week is calling BS on the whole discussion. I actually, you know, try very hard to explain that there is no such thing as AI, as an independent actor. And therefore, the entire scare is misconstrued.

00:04:24 Andrew Keen: Yeah. But we're not again, I keep on saying this. We know your position. We've been over this week after week. You're not sympathetic. And in some ways, I agree with you.

00:04:32 Keith Teare: Sympathy. Andrew, it's not about sympathy. It's not about opinion. AI is not dangerous, period.

00:04:38 Andrew Keen: Yeah. But we're not but that's not my question. The question is, are we in a new political reality in your sense? I mean, the fact that the three major newspapers are all leading with the same story.

00:04:48 Keith Teare: I think we are, but it represents, a strong economic self interest by the AI leaders to contain open source AI in its growth. So it is a it isn't that they are bowing down to a mass movement. They have an extremely strong self interest in putting rules in place that prevent others from playing.

00:05:20 Andrew Keen: But coming back to the politics of this, what's your take on again, Congress is not made up mostly of particularly sophisticated people, but they are sophisticated in when it comes to technology. But they are they have good senses when it comes to politics. And we know and I mean, I'm not knowledgeable about AI to make any comment on these doomsday threats. Some of them do seem very exaggerated. But is Congress waking up to it? I mean, waking up might may not be the right word. Suddenly recognizing that, for them, at least, it might also represent a political opportunity. The only person who's who doesn't seem to be too bothered about it is Trump. He was asked about, AI's existential risks. He said, I don't have any, although he seems increasingly marginal as well in American politics these days.

00:06:17 Keith Teare: I thought Scott Bessent's speech on this issue is probably the most prescient speech of the week. Bessent, to an audience made the point that, one, it's impossible to stop what is called recursive self improvement. AI, creating, better AI. It's impossible to stop it even if you could get OpenAI and Anthropic to, in quote, slow down, that wouldn't slow it down. Because there's now, in the Hugging Face store, there's 3,500 AI models, mostly open source. So the idea that you can stop it or slow it down is mistaken. And therefore, Congress, ultimately, will conclude that they can't punish America at the expense of the rest of the world by enforcing a slowdown. They won't be able to achieve that. It will be politically, very, very silly of them. And so I—

00:07:26 Andrew Keen: But they've done many silly things in the past. I certainly wouldn't put it beyond, for better or worse, again, Congress. So you're not really answering the question, though. Do you think we're in new territory when it comes to this conversation? Maybe not so much this weekend, but over the last few days.

00:07:44 Keith Teare: I look. I don't think the fact that last week's, title was don't trust the trust scare tells us there's nothing much new in the last week. If anything, we're a little bit ahead of the curve. But go and look at the titles of That Was the Week for the last six months.

00:08:00 Andrew Keen: Yeah. Because you keep on repeating the same thing, but that's not my question. We know where you stand.

00:08:05 Keith Teare: Your question was, is there anything new? And the answer is no.

00:08:08 Andrew Keen: No. But is there anything new in political terms and in terms of, you know, I began with—

00:08:13 Keith Teare: It's been that the heat is ramped up a notch, Andrew. That's about it.

00:08:20 Andrew Keen: Well, it was an interesting piece by Stephen Witt in the New York Times. He's come out being concerned too. He wrote The Thinking Machine. He's certainly not anti tech. He wrote the sort of authoritative book about Nvidia last year. He talks about four ways of addressing this crisis or what at least he sees as a crisis. Shutting it all down now, which, might be a little over the top. Second, take an air crash investigator approach. Thirdly, monitor the situation. And fourthly, flip the kill switch. Are any of those in your view viable? I mean, at the end of his piece, which I thought was quite good, he says that the more, and I'm quoting him here, the more I learn about these rogue systems, the more I'm reminded of the first two months of 2020, when a small group of epidemiologists were attempting to raise alarms about COVID. So at least in his mind, we've been through this before. Are any of those four? I mean, I know you're I we know where you stand on this, but are any of them credible?

00:09:33 Keith Teare: Look. I think from a logical point of view, if you believe and this is what Dario says in the essay he wrote today or he published today. I don't know when he wrote it. If you believe that repetitive, that sorry. Recursive, self learning cannot be controlled by humans because humans can't be fast enough to see all the variables taking place, and if you believe that the AI agents produced by, this self learning, will be malicious in some way, the only logical thing is not to slow it down, but to shut it down.

00:10:19 Andrew Keen: So—

00:10:19 Keith Teare: If you truly believe that humanity is threatened and that humans can't control it, you would have to shut it down, which means Anthropic should cease existing as a business. There should be no IPO. You would ban AI period because it's so dangerous. The idea that it's out of control, but, hey. Guess what? We're clever. We could slow it down. It doesn't make any sense. You either control it or you don't.

00:10:49 Andrew Keen: Yeah. And I think with Witt, he's got this I don't know what the difference is between shutting it all down now and flip the kill switch. I'm not sure what they—

00:10:58 Keith Teare: Yeah. They feel like the same to me. And but, you know, all three of the leaders are buying into the idea that slowing down is possible. Now, you know, recursive self learning is binary. It either exists or it doesn't. The minute there is recursive self learning, there's no speedometer. It's either on or off.

00:11:30 Andrew Keen: So you're saying that none of these options that Witt lays out from shutting it all down now. What about the air crash investigator approach? I'm not sure if you've read his piece. Yeah.

00:11:45 Keith Teare: I have read his piece. I mean, look. The he's a smart guy, and none of the things he talks about, are stupid unless you ask, is he solving the problem? If the question is he solving the problem, the then— and you agree there is a problem, which I don't, but let's pretend there is a problem. Shutting it down is the only option.

00:12:08 Andrew Keen: Yeah. And I think that, as I said, we know where you stand. I'm not equipped to understand it enough. I mean, I trust people like Witt, I guess, although I respect what you're saying as well and what Bessent is saying. I think it's more complicated, though, because the reason why Trump doesn't have any issues on it is because he knows if they shut it down or slowed it down, it had a dramatic impact on these upcoming Anthropic and OpenAI IPOs in which the increasingly, the American economy is dependent on. So it's not just political realities, but it's economic realities too, which complicate the issue.

00:12:52 Keith Teare: Yeah. I look. As we know from Trump, he sees everything through, a national framing.

00:12:58 Andrew Keen: Well, from his own self interest. I don't know.

00:13:00 Keith Teare: Well, let's just for a second ignore his self interest. That we can take it for granted. He always has self interest. But I actually think his, the what he shares in common with, Andreessen Horowitz, this American framing means that, it's impossible, if you accept that there's an AI race, which again, I don't. I think it's a global thing and there's no endgame. It just keeps going on. But if you accept there's a race and you want America to win, you can't penalize American AI when the rest of the world is racing ahead. So I think he has a kind of a logical national framing for this whole thing, which makes him an enemy of slowing down. Plus—

00:13:53 Andrew Keen: But there is also the real the if not, I don't know if it's the political or the economic reality of these upcoming IPOs and the economy's broader dependence on it and the tens, hundreds of billions of dollars being poured into this sector. I mean, no one's gonna be shut down. I mean, the only way it will get shut down is if there's some sort of 9/11 event or maybe a COVID event. But, otherwise, it's not gonna be shut down.

00:14:21 Keith Teare: Well, you know, I think there is increasingly and Bernie Sanders is leading this. There's an increasing opinion that it should be stopped. They're using the word stopped. Paused is the softer version of the word, but it all amounts to discontinuing, closing down. Whether it reopens again on some new basis—

00:14:46 Andrew Keen: But I'm not showing much power, Bernie. I mean, Bernie Sanders isn't even powerful within the Democratic Party, and they—

00:14:51 Keith Teare: No. But it no. But it's a growing opinion. And that, you know, it's one of those be careful what you ask for moments. I personally think most innovations only become acceptable to those most impacted by them after they've succeeded. In other words, once they deliver the bacon. An AI probably is no different to that. Most people have to experience a positive impact on their life from it to discount some of these fears.

00:15:27 Andrew Keen: Right. One of the pieces this week was about how, when will average people feel AI's impact for better or worse? I mean, in a sense, they are already because the economy is heating up. Again, I'm not sure whether all average people are well, I'm not entirely sure what an average person is, but there is a stock market. The strength of the stock market is in many ways still being driven by tech and these impending IPOs. So the story certainly hasn't ended. It's becoming much more interesting, I guess, in political and historic terms.

00:16:02 Keith Teare: Yeah. I think if Trump tied his $5,000 for each American adult to the GDP growth driven by data centers, giving the government, money in the bank, which, of course, is only part of it. The rest is tariffs. He could turn, you know, he could turn AI from being a net negative to being a net positive. It's unlikely he's gonna do that.

00:16:28 Andrew Keen: I somehow suspect that Trump has less and less influence on the national conversation, but we will see on that. Not everyone agrees with you that, there isn't an AI race. Noah Smith, who you often choose, one of the most popular economic writers on Substack, I think he's one of the one or the first or the second most popular person on Substack, has a piece which you link to America's still beating China in the AI race. So I take your point that it's still a global technological innovation, but one can quantify who's doing better than the other. Can't one?

00:17:06 Keith Teare: Well, you can if you under if you know how the race will end up being won, in quotes. But, really, the China America debate is an open source versus centralized closed source debate.

00:17:23 Andrew Keen: And, ironically, of course, it's the Chinese who are in the open source camp, whereas the American are in the closed source—

00:17:31 Keith Teare: Correct.

00:17:32 Andrew Keen: Which is particularly ironic and troubling, I think, for some Americans.

00:17:37 Keith Teare: Yeah. I mean, historically, open source leaks like water into every crevice, and, surrounds everything and eventually strangle it. That, you know, data centers are full of, computers running Linux, not Microsoft Windows or Apple, macOS. The cloud is almost 100% Linux. So open source has a habit of getting everywhere.

00:18:04 Andrew Keen: I mean, to put it I mean, I know you believe that some of this call for this AI slowdown is the self interest of the hyperscalers, Altman, Musk, and Amodei. But if they were to slow down, they're not gonna convince the Chinese companies to slow down. So in some ways, if US AI slows down, it's a huge victory for Chinese open source, wouldn't it?

00:18:32 Keith Teare: Potentially. There's a lot of unknowns about China. Like, yesterday was a revelation, and I still haven't figured out if it's true or not that Kimi was sending every query to Claude, and therefore, there was actually no Kimi.

00:18:51 Andrew Keen: Mhmm.

00:18:53 Keith Teare: And it is the case trained using inference on the American models and therefore are always about a year behind. So there's probably a dependency chain there that doesn't go away. And so, realistically, the idea that China would go ahead if we slowed down it, I think it might not be true. I think there'd always be a gap.

00:19:19 Andrew Keen: Yeah. My guess is that it will take a major disaster of some sort rather to bring the Americans and the Chinese to the same table. Alex Kantrowitz always seems to be quite sensible. He has a piece, it's time to bring some sanity to the AI risk conversation. Now I take your point that you don't believe that AI offers an existential risk. But, Keith, given that so many people do, is there a conversation, a sane conversation, which everyone can come to the table and recognize Yeah. The legitimacy of each other's positions?

00:19:57 Keith Teare: Well, there's a sane conversation. You know, the thing about politics is it doesn't matter whether you think the other side is wrong. Your job is, to win the argument if you disagree with them, and you won't win the argument by shouting at them. So there's always the need for a sane conversation. Unfortunately, the form that's taking, and it may be insincere, is the AI leaders pretending that they share the concerns and putting in place palliatives, which if the critics are right, won't solve the problem.

00:20:36 Andrew Keen: Why are they doing this? Just out of self in your view, out of self interest?

00:20:42 Keith Teare: There's many theories of it. I guess the most benign theory would be they want to keep developing. And so they have to, you know, a bit like in England, we all always put in place a royal commission whenever the people get angry about something. And the commission, like, you know, looks at the evidence and sometimes takes years to arrive at conclusions. They're saying they're gonna put in place external monitors that are working inside the company and so on. There's all these kind of plans to assure the public that safety is being is steps are being taken. But look, if recursive self learning is teaching agents that can't be relied on to follow rules, nothing you do to slow it down or monitor it is gonna be helpful.

00:21:39 Andrew Keen: So in a way, you're kind of calling for shutting it down. If you're wrong on this recursive issue, if you're wrong that AIs can't think for themselves, then the whole thing should be shut down. Are you saying that?

00:21:53 Keith Teare: No. I'm saying that's where it that's the only conclusion you could arrive at to satisfy the other side. And if the other side accepted anything less than that, then you have to doubt that they believe their own arguments.

00:22:08 Andrew Keen: So are you saying then that the Altman and Musk and Amodei, they don't believe what they're saying? So these headlines today, it's just some scam on their part and that there's another agenda for them.

00:22:20 Keith Teare: Well, I said correct, but actually, let me just think about that. I usually, people are sincere. So they're either very naive. They think that this reaction will take the sting out of the or the heat out of the conversation and give them some space, or they're manipulative. So let's take the first that they're just naive. It won't end in a good place. The demands to shut it down will get louder the more capable the AI becomes. So unless you can show people that there are benefits to AI and that the safety issues are not real. Of course, always put in place whatever controls you can. I, you know, I think agents, going, rogue in the lab is easy to control. The fact that they said they can slow it down, by limiting the resources made available to any given agent, that's real. They can do all of those things. But, ultimately, if an agent is malicious and powerful, you've got to assume it will work around whatever limits you put in place. I disagree with that premise. I don't think agents are conscious, thoughtful, goal orientated.

00:23:42 Andrew Keen: Yeah. I think you're increasingly in a small minority here. I mean, who's in your camp in the tech community now?

00:23:51 Keith Teare: I'd say pretty much everyone.

00:23:54 Andrew Keen: Well, I mean, Amodei, Musk and Musk, Amodei and Altman aren't in your camp.

00:24:02 Keith Teare: I think that Musk and Altman are, they're just, being politicians. I think Amodei might be as well. I mean, remember, he's running Anthropic, which is racing to improve AI and do a—

00:24:18 Andrew Keen: Yeah. But he wants to slow down.

00:24:20 Keith Teare: And do a $2 trillion IPO. He well, he wants to slow it down for what reason, Andrew? What's his reason?

00:24:26 Andrew Keen: Well, I mean, what he says is it needs to be slowed down, that and I'm quoting from the headline in the FT that humans could lose control of technology. So—

00:24:38 Keith Teare: I think Altman called his bluff this morning. Altman said in his interview with Fortune that OpenAI will not be doing an IPO this year because given the discussion around safety, it doesn't appear to be the best moment to do it. Well, what about Dario? Is he gonna pull his IPO?

00:24:56 Andrew Keen: Yeah. Now that's a good question.

00:24:58 Keith Teare: I don't think he is. I think he's playing a game.

00:25:02 Andrew Keen: Well, I—

00:25:03 Keith Teare: High stakes game, but he's playing a game.

00:25:07 Andrew Keen: Well, we will see. As I said, I think we are in my sense is we're in new territory. You haven't really said that.

00:25:19 Keith Teare: Well, I kind of get your point when it comes to—

00:25:22 Andrew Keen: I mean, the public or whatever the technological truths or lack of truths, there are political truths and realities and economic ones. And it seems to me that we are, as I said, edging into new territory when both the left and the right seem to agree on this. No. It's Congress. Both Republicans and Democrats seem to be, quote, unquote, waking up. They always seem to be asleep, for better or worse. But we will see. I mean, it's certainly not gonna change everything this weekend. This will be a subject we come back to in our show.

00:25:56 Keith Teare: But do you if you roll the clock forward to after the midterms, I think, the US government, the White House level does nothing to regulate AI. It just won't happen. Congress, it's an open question how it turns out, Congress. I think it's pretty hard for the Democrats to get a majority of both houses, but they could. I think that's gonna be what the real change is. And if the Democrats don't get control of both houses, my guess is nothing changes. Then, you know, the market settled down. The IPOs probably do happen, and AI continues to be developed.

00:26:39 Andrew Keen: Well, that's the best case scenario, and that's if there is no further disaster of one kind or another. That's if you're right on this—

00:26:48 Keith Teare: Well, when we say further disaster, but there hasn't been any disaster so far.

00:26:53 Andrew Keen: Well, if there is a disaster, so that's all speculative. Let's move on. We've talked about this enough. You had an interesting piece you connected. You sent to me about what writers who use AI want you to know. We you and I both use AI. I think it's hard to imagine any writer who wouldn't use it, although there are some. What does this piece tell us about how AI can help writers?

00:27:21 Keith Teare: So this was written on Every, which is a kind of a aggregated publication of good writers in general. And it was interviews done by the authoress with various writers, one of which was, Kevin Roose.

00:27:39 Andrew Keen: Roose.

00:27:40 Keith Teare: Or Roose. And, it was quite detailed, process, disclosure about how they use AI. And to be honest, it was quite similar to the conversations you and I have had. Kevin Roose has written a book, 428 pages [ed.: The AGI Chronicles is listed at 448 pages], which they used AI extensively for. He says that he created the chapters, and the structure of the book. And then AI did about three years worth of work, from that starting point, which is quite similar to what I did with The Human Dividend.

00:28:21 Andrew Keen: Well, yours is already a bestseller, isn't it? How many millions of copies have you sold over or given away?

00:28:27 Keith Teare: Yeah. Yeah. You know, you go for these low blows.

00:28:31 Andrew Keen: Yeah. But you said it's a book. I mean, Roose spent four years on it. You spent what? Four weeks—

00:28:36 Keith Teare: No. He didn't. He spent one year.

00:28:38 Andrew Keen: Well, he spent a year.

00:28:39 Keith Teare: And then he says it's four years worth of effort because of AI. Yeah. Because [mine] isn't published yet, and I've so far spent three months. So maybe it'll be a year by the time I publish it.

00:28:48 Andrew Keen: Well, I have to get. But, so yeah.

00:28:51 Keith Teare: I— But you shouldn't be so cynical because, actually, I think what Roose discloses there as a process is exactly the process I used.

00:29:00 Andrew Keen: Well, I mean, it's a process that many writers will use in terms of crunching information. I'm not sure. Did Roose say he used AI in terms of the actual writing or just to sort of manage the information?

00:29:12 Keith Teare: No. He used it, and then he went and edited it. He kicked it off with chapters, which he says it took him most of the six months of his work was figuring out the structure of the book. He talks about it in quite a lot of detail. Once he had the structure and the main narrative and the arguments, which were bullet point style and, chapters with bullet points, then AI kicked in, and then he edited it. And he said it the whole thing took a year, whereas it would have taken three to four years if he did all the work. Yeah. I think that's okay. I mean, I think that's a very good process.

00:29:48 Andrew Keen: Yeah. I don't think anyone would argue on that. There was a good piece, in the Financial Times. I was really struck with a piece on moonshot capitalism, AI rewrites the venture capital playbook. And it really was astonishing how much capital is being poured into this moonshot sector and particularly, how America continues to dominate. One of the images is of US, VC funding on deep tech at, two I guess it was at $246 billion, China at $105 billion, and then the next country after that is the UK at 26. So it really is astonishing. I know you don't always like to think of these things in national competition terms, But it really is astonishing how far the US is ahead on this.

00:30:47 Keith Teare: Oh, yeah. Well, the US is—

00:30:49 Andrew Keen: Especially compared to Europe. I mean, Germany has only put $11 billion in. I mean, that's, Israel's done more than Germany. Israel's at $12 billion, Sweden at 13, France at 14, UK at 26, and the US at 246.

00:31:06 Keith Teare: And the US is really 80% California. So if you think of it that way, California is that much bigger than any European nation. No. The look. The US is from a financial point of view, the entire history of the US with Wall Street, private equity, and venture capital collectively has created the possibility of huge and rapid real time investment in opportunities as they arise without any delay because there's no policy oversight that stops it. China, there's a huge policy oversight. Despite that, it's second. And in Europe, there's a regulatory framework that's about 10 times slower than China's.

00:31:53 Andrew Keen: Yeah. I mean, the European numbers are, astonishing in terms of the comparison.

00:31:59 Keith Teare: So you could think of that as regulation inhibit inhibited Europe, China somewhat, but not enough to stop them being second. And the US, there's none at all, which is why there's such a strong voice in Silicon Valley against regulation because—

00:32:16 Andrew Keen: Although there is some. I mean, you have a piece this week about California enacting laws restricting chatbots and banning teens from, quote, unquote, addictive social media. So, I mean, there is some regulation in the US. It's not just a wild west.

00:32:34 Keith Teare: Yeah. I think it's very light and not very well policed, in general. I mean, have you noticed any teens stopping using social media in the last week? I definitely haven't.

00:32:46 Andrew Keen: Well, I don't know any teens. My kids are older.

00:32:49 Keith Teare: Yeah. Mine too. So may maybe we're not typical, but I, you know, I do think the point in that article, which is really important, is the venture capital power law, which is roughly the law that says 80% of the gains will come from, you know, 2% of the companies, means that money is piling in to a very small number of companies at great scale. And that is, to be read alongside Andreessen Horowitz published something yesterday saying that returns from venture capital after the SpaceX IPO now dwarf returns from private equity, which historically has been much bigger than venture capital.

00:33:34 Andrew Keen: And that's why this whole issue of slowing down AI or shutting it down or flipping the kill switch, That's why it's much, much bigger than just a technology issue. It's a it's an issue that addresses the whole future of the American economy, which explains Trump, maybe explains in some ways Amodei, but it makes it enormously consequential in historical terms. I'm not sure how it's gonna work out. Yeah. It's gonna be very interesting.

00:34:03 Keith Teare: I don't think money would stop a shutdown of AI if AI threatened humanity in the way that's being implied. I mean, honestly No.

00:34:15 Andrew Keen: I mean, no one's saying that.

00:34:17 Keith Teare: Right. So if any of us don't—

00:34:19 Andrew Keen: know the threat I mean, if, I mean, if you're not sure, but you suspect it could destroy humanity or create some disaster, then, maybe the economic downside of shutting it all down plays a role in your decision.

00:34:37 Keith Teare: Well, if you think of the chain of events that would have to happen, it's actually a really good exercise to gameplay. What would actually happen for AI to threaten humanity? Probably the only two weapons that could do that are biological or nuclear. So what would be the steps an AI would have to take to get, the ability to both develop, deploy, and successfully land a biological or a nuclear weapons, that would, erase humanity. The sequence of events is fantastical.

00:35:20 Andrew Keen: Well, whatever. Yeah. It's like 9/11. You would've, might've said— we've just celebrated—

00:35:24 Keith Teare: 9/11 is one airplane. The this is way more complex than that.

00:35:29 Andrew Keen: Wow. But if you'd have said, oh, these guys are gonna hijack a plane or hijack a series of planes and crush them into the World Trade Center and destroy all the buildings and crash into Pentagon. You say, well, that's not very realistic. That's not gonna happen. Anyway, let's move on. We know what you think on this. One good piece of news I thought was your company of the week or startup of the week is Starbucks, who don't we don't think of as a high-tech company. They bet a billion dollars on coffee on a coffee house antidote to lonely digital lives, which I think speaks to the economic opportunities to some of the downsides of tech. Do you think that this might help the loneliness epidemic, or is it just a clever marketing scheme on the part of Starbucks?

00:36:24 Keith Teare: My camera changed, Andrew. I think the main one—

00:36:27 Andrew Keen: Yeah. I think you've been hacked by Hugging Face or by Dario because you've been so rude about it.

00:36:32 Keith Teare: Exactly. Yeah. I thought it was interesting, and it is kind of a startup. Starbucks is turning into do you remember Barnes & Noble's sofas and tables? Yeah.

00:36:41 Andrew Keen: They didn't do very well, so I'm not sure that.

00:36:44 Keith Teare: Well, they did for a while. For a while, it worked. It— People came in. I mean, with Starbucks, there is a nightmare scenario where, you know, San Francisco's homeless all go and lays around on sofas in Starbucks. But they are kind of reinventing—

00:37:01 Andrew Keen: [Is] that a nightmare for— The Starbucks or the homeless people? I'm not sure you'd call it a nightmare. That might be a bit of an exaggeration. That's the Palo Alto view.

00:37:12 Keith Teare: Yeah. Anyway, look, they're reinventing themselves as a place to hang out. And that it's counter the, atomized individualized culture of digit— the digital world. And they're making a bet on people, you know, hanging out in a kind of, community like chat space, a real live chat space. I think I bet against it working. You know, if I was analyzing Starbucks as an analyst, I think the biggest problem is the coffee isn't very good. So you don't go there. So they—

00:37:48 Andrew Keen: I don't drink coffee. I sometimes go there because, it can be a pleasant place to sit if you—

00:37:53 Keith Teare: If you need a place to sit, yeah, with your laptop.

00:37:56 Andrew Keen: But I think what it does speak to, which is, I think, fairly obvious, is that, there are many analog opportunities in the increasing digitalization of life. I mean, I know some kids doing startups around this loneliness epidemic too. So that clearly it speaks to the economic opportunities of some of the problems with tech, and I'm sure there'll be many when it comes to AI. My interview of the week is with, Alex Edmans, popular London Business School, writer, professor. He has a new book out, The Madness of Markets, Why Smart Investors Make Crazy Decisions. His, example is Isaac Newton, who lost a fortune on the South Sea Bubble. Could also use the example of Mark Twain. This came up in a show a couple of weeks ago with Jeff Jarvis, who lost a fortune investing in the wrong typeset technology in the late nineteenth century. You're an investment professional, Keith. Do you agree with Edmans? Do you think that really smart people like Newton and, Twain make just as dumb decisions as anyone else?

00:39:19 Keith Teare: Yeah. There's a long history of dumb decisions. I remember when I was starting my first Internet company in 1994, EasyNet, the craze at that time was interactive television. And the, cable networks and the broadcast networks were spending billions and billions and billions of dollars on making your TV interactive. And my sociologist background, I had a very simple pushback, which was the living room usually is for more than one person. And if the TV becomes interactive, whoever is changing it is gonna piss off the other people in the room. And so by definition, TV is not intended to be interactive. It's a social viewing medium, not the private—

00:40:08 Andrew Keen: No. But his point is that smart people think they're smart, so they feel they can just you know, that they have an instinct. I mean, I don't know who we consider really smart people. I mean, Einstein isn't around these days, but the equivalent of an Einstein today. Would you trust them? Do you pitch them for your investment platform? I mean, who— or—

00:40:32 Keith Teare: Let me revise the question. You're talking about experts, really, aren't you? And I'd say experts—

00:40:37 Andrew Keen: I mean, people we consider smart. I mean, who in your mind what type of person makes the best investments? Is it a—

00:40:44 Keith Teare: The well, weirdly, the history of Wall Street tells you the best investors are working class street smart kids who become brokers. It's not experts. It's people who smell a deal.

00:41:00 Andrew Keen: And recognize I mean, Edmans' point is they recognize their own limitations.

00:41:05 Keith Teare: I don't even know if that's true. I just think that inventors are not gamblers. And so almost all inventors will be a bad investor. An inventor equals smart equals, you know, clever. I wouldn't expect them to be good investors. Now there are exceptions like Elon Musk. He's not actually an inventor, Elon. He's a more of an operator, but he's a very good investor, you know, and has shown that. I doubt that Altman would be a good investor. I don't think Dario would be a good investor. Yeah.

00:41:44 Andrew Keen: I think Dario would be good at anything.

00:41:47 Keith Teare: I think he's pretty good at politics. He's pretty good at politics.

00:41:51 Andrew Keen: If we were doing this show five or ten years ago, Keith, we wouldn't have talked about any of this. We would have led and spent hours talking about, the Apple announcements this week, the iPhone Duo, the first, iPhone, split screen device. Is that because we're in a new moment, or is it because AI is so dominant? You linked to a Ben Thompson piece that suggests that in the context of this Apple announcement this week, they're still missing with their new CEO. They're still missing an AI strategy, which I actually agree with. So why have we left it right to the end to talk about Apple? Does it not really matter anymore hardware?

00:42:43 Keith Teare: I think it can, but I don't think any of their announcements had a big enough impact on the primary trends. I mean, the foldable iPhone, I'm not getting one.

00:42:56 Andrew Keen: Yeah. Because you don't like foldables, but they say that the foldable every foldable made so far has been kind of a piece of crap, and they thought it through. And they and the early reviews have actually been quite good.

00:43:08 Keith Teare: Are you gonna get one?

00:43:09 Andrew Keen: I'm open to looking at it. I'll certainly go and have a look.

00:43:12 Keith Teare: I mean, I did buy the, Android 10 foldable sorry, nine foldable. And it was very similar, very beautiful screen when you opened it up, and it had four screens altogether, four sides. And it works. I just, you know, I don't think I have a need. I don't think it serves a need for me. I've got an iPad, and I've got a phone. If it replaced both, maybe I'd think about it, but it's too big. It's a weird shape.

00:43:43 Andrew Keen: We haven't seen you haven't physically seen it yet. Have you ever seen your hands? [as spoken]

00:43:47 Keith Teare: Well, I've seen humans with hands holding it so I can see it.

00:43:50 Andrew Keen: Well, we'll see. I mean, do you agree with Thompson that Apple's still basically missing an AI strategy? It's still amazing that they even use the word Siri. I mean, I would have thought that would have been eliminated, Yeah. And they would have come up with a new concept, a new name, a new a set of languages to describe what they're trying to do.

00:44:09 Keith Teare: Well, it's the same as Amazon. They still have Alexa.

00:44:13 Andrew Keen: Yeah. But Amazon, it's not necessary. Alexa's not core to Amazon's business, whereas Siri and AI for Apple is—

00:44:21 Keith Teare: Look. Look. I will say this. I've been running the new OS for since the developer conference in June. It is way better, and, it does more. And it doesn't rely on canned answers like Siri used to. It actually can, you know, think, in quotes. So it is a lot better. What Apple is lacking is what, weirdly enough, Facebook did this week. Facebook released Muse. Muse is their version of OpenClaw, or Grok Bot, which is a full agent with a virtual machine able to work on any task, including accessing all the files on your laptop or your desktop. Very powerful, and is the probably the primitive, that is to say, the base case of what the future looks like, where everyone on Earth has access to, in quote, superintelligence that is isolated from everyone else's and serves only them. That appears to be where everything's going. OpenAI offers that now. Anthropic offers that. Grok Bot offers it. Now Facebook offers it. It's standardized. And so Apple doesn't offer that. Apple is bolting on AI use cases to existing apps and stops short of giving you an agent that you control.

00:45:52 Andrew Keen: Yeah. It comes back to your conversation about open source and, of course, ironically, or for better or worse, Facebook. Most people don't trust Facebook, certainly Zuckerberg. But they are committed to open source because they're not in the Altman, Musk, Amodei camps. It's interesting that I'm sure Mark Zuckerberg isn't gonna be calling for an AI slowdown, or maybe he's calling for a slowdown on them, but not on open source. So yeah. And, of course, Apple has always been a walled garden. They're dependent on that, so they're not gonna be sympathetic.

00:46:25 Keith Teare: And the scale required I mean, imagine Meta has roughly 2.7 billion unique users in a month. If each one gets a virtual machine, even if you can put a hundred on a single physical machine.

00:46:40 Andrew Keen: 100 what?

00:46:41 Keith Teare: A hundred people. A hundred virtual machines on a single physical machine. You know, a virtual machine isn't an actual piece of hardware. It's a software defined machine. So you can put, let's say, a hundred of them on one physical computer if it's a big enough computer. Even doing that, you get from 2.7 billion, you reduce it by take one zero off, you get 270 million. And take another zero off, you get 27 million. They would need 27 million physical machines to support Muse. That's huge.

00:47:20 Andrew Keen: Well, we will see much to come, much of these discussions. What isn't to come is any more Bill Draper news. He died this week, one of the fathers of venture capital in Silicon Valley. You have, a post from his son, Tim Draper, who's been on the show. You and I know him. He was an investor in one of your companies. Was he a remarkable figure, Draper, or just another VC, Bill Draper?

00:47:50 Keith Teare: Well, he was one of the first VCs, if not the first. He's probably most well known because he was the first investor in Skype, before his, son, Tim, went and invested in Skype. But he was very early in that Lockheed Martin, Silicon Valley emerging era and was a real venture capitalist in the same way that Tim is. He sought difference and contrarianism that could turn into big outcomes that no one believed in at the point of their formation a bit like AI now. So, yeah, he was, a very important figure, and everyone knows him because he's been around for decades doing that kind of work, in the most recent period out of his family office.

00:48:40 Andrew Keen: Because the Drapers now are a [unclear] of VCs. Is there a danger of the VCs? I mean, you can't blame Bill Draper for giving his son and his grandkids opportunities. Is there a danger? Will we have a an Andreessen family lineage or a Ben Horowitz? Is that the future of VCs?

00:49:01 Keith Teare: I don't think so. But Tim does have a son who's a VC, who's very good as well.

00:49:05 Andrew Keen: And a daughter.

00:49:07 Keith Teare: Adam. Yeah. He's got more than one child, but Adam has trailblazed as a VC, and he's doing very well. Look, you can have your opinions about anyone, but I think there's no doubt that on the week of his death, we should recognize that Bill Draper had a big impact in what we now experience in Silicon Valley.

00:49:29 Andrew Keen: Well, there you have it. It might be a historic week. It might not be. It was the week where at least Altman, Musk, and Amodei agreed for a call to an AI slowdown. Keith doesn't believe in the slowdown. He's not convinced it's a major historical moment. I'm not sure what that particular announcement is, but I suspect that the die is cast now in terms of what will happen with AI. I'm not sure exactly what will happen, but something will. And we will be following it, Keith, in the weeks and months to come. So have a good week, and we will talk next week.

00:50:04 Keith Teare: Adios.