Speaker
[music]
Words and timings
No per-word timings found.
Transcript Viewer
Aug 2, 2026 ยท 2026 #28. Read the transcript grouped by speaker, inspect word-level timecodes, and optionally turn subtitles on for direct video playback
Speaker Labels
Edit labels for this show, save them in this browser, or download a JSON override for the production folder.
Transcript Playback
Human Transcript
Blocks are grouped by speaker for readability. Expand a block to inspect word-level timing.
Speaker
[music]
No per-word timings found.
Andrew Keen
Hello, everybody. Happy August. We finally arrived in the summer of 2026. A couple of weeks ago, we did a That Was The Week tech summary show with my friend Keith Teare, the publisher of That Was The Week, entitled Who Owns Intelligence? We talked about the smart wealth of nations and Keith, I don't know if it was an acknowledgement or a confession or a boast, he told me that he is in the process of writing an AI helped book. I'm not sure entirely AI authored, but certainly involved with AI, appropriately entitled Who Owns Intelligence? So Who writes books in our age of AI? And should authors use AI? That's a subject that we are talking about once again this week. Lots of stories this week. There's a red hot book at the center of an AI mystery. That The Wall Street Journal writes about that's taking over the publishing industry. I assume it's not Keith's. Snapchat has joined other popular platforms in its fight against AI slop. It's joined YouTube, LinkedIn and Substack to combat what it calls fake writing, images and video. And even the Financial Times is getting into the business. They had an interesting piece this week about the new premium product being books written by people, as in contrast, I'm guessing, books written by AI. Keith has a different take on it, though. This week in his That Was The Week editorial. In an editorial entitled AI Detected, he says it's not cheating. You should use it at will. Keith, in terms of using it at will, what does that make of your Who Owns Intelligence book? Who's going to buy a book in publishing if it's been written by AI?
Keith Teare
I think the opposite might be true. I think it might sell more. Because it's of the times and it's going to be intriguing to people. But before I provoke you with that thought, let's just stand back a little bit. Nothing is written by AI. There is absolutely nothing in the world written by AI because AI doesn't have any will at all. It won't even write a single word unless you ask it to. And so... Let's just start there. Intelligence is the word we use for the collective wisdom of the human race and dumb things as well, not just wisdom. It's our collective evolved understanding of ourselves, the world, and so on. So it's only human. And when a human uses AI to do anything, they are leveraging what they know in order to produce some kind of an output And AI is no different than any other tool they might use. If you go way back, and I say this in the editorial, in platonic times, the very idea of writing something down was considered bad because in those days they believed that an idea could only live in a person's brain and come out of a person's mouth. And the minute you wrote it down, It deteriorated in value.
Andrew Keen
Well, yeah, I mean, I take your point and you're right in a way, although of course not everyone agreed on that. I mean, Socrates believed that, but others didn't. So there was a debate back then as there is today about how much tech we should and shouldn't use when it comes to creativity.
Keith Teare
Yeah, but the short story is I think it's already strange for an intelligent person To produce any creative work, not just books, and not engage with AI as part of what they do. I mean, why wouldn't you? I'm not saying there's anything wrong with not using it. If you don't want to use it, great. You know, I sometimes buy things on Etsy that are craft made by human hands, but most of the stuff I buy isn't. It's made in factories. And there's no value judgment between the two. One isn't better than the other. And writing with AI isn't worse than not writing with AI. It's just a tool.
Andrew Keen
So how, and you, we've had this conversation before. You asked me for some advice about finding an agent for your Who Owns Intelligence book.
Keith Teare
Coining a phrase, finding an agent.
Andrew Keen
Right, finding, exactly, finding a human agent as opposed to an AI. I'm not sure how serious you were about that. I'm not even sure how serious you are about this book. You have a day job, which takes up most of your time, and then you have your newsletter and this show. But what is a publisher supposed to do if they believe, rightly or wrongly, that AI mostly wrote this book? Most agents, literary agents these days, I think, would say, well, we don't want to go anywhere near it. Are they wrong?
Keith Teare
Completely wrong. They're going to lose their jobs soon. I think that it won't matter. It literally won't matter. Either way, by the way, it won't matter. It really all depends on the quality of the ideas and the thinking if it's a nonfiction book and the quality of the story if it's fictional. And there's nothing about using AI that makes it worse. And it may make it better. So I think the whole discussion is a little bit strange. It's understandable, but it's a little bit strange.
Andrew Keen
So you're saying then, for example, that this debut novel by a writer called Jerry Falarde that was fought over by 14 publishers, And then they discovered that he, I don't know, again, I'm using this idea of him using AI, and I don't know how or when he used it, that these publishers and the agent are wrong to retract their support and interest in this product.
Keith Teare
Right. Well, the irony is the very last sentence of that story from the agent who's dumped the author says it's a fantastic book. And so what is driving this schism actually is lack of transparency. I think lack of transparency is a much bigger problem than using AI. Using AI and not and being shy about it because you fear offending somebody. And therefore you end up looking as if you're cheating or being devious. That's the problem. You should just own it. Own the fact that you use AI. And when I wrote that in my editorial, I went to that book that you pointed to and I put a strap line underneath my name that said something like AI was heavily used in the writing of this book. Because I think it's really more about transparency. I actually think every book's going to have that soon because who won't use AI in some way or other in writing?
Andrew Keen
Yeah, it's interesting. I mean, with the invention of word processes and computers, no one ever came up with that caveat for books that they were written by hand or on computer. I wonder... How realistic it is because no one's going to be entirely honest. Some people will have a little bit of guilt saying, well, I couldn't have written this without AI or this sentence required a bit of sharpening up. This wasn't really mine. Are we doing away with the idea of human originality? We talked about that last week and the unintended legacy, the ironic legacy of French post-structuralism. That we're no longer quite as authoritative as we thought we were.
Keith Teare
Here's the book that I wrote in 19... I wrote it in probably 1986. It was published in 88 with a very famous preface by John Pilger, who has now passed away.
Andrew Keen
He may have passed away having written it.
Keith Teare
Well, this book was edited by James Woodhausen and sub-edited by Mick Hume. And every book has an editor and a sub-editor. In fact, a publisher won't even look at your book unless you have an editor and a sub-editor. So how much of a book is really the author's work anyway? It's ridiculous to say that AI changes authorship, because if you're going to do that, you should say that editing and sub-editing changes authorship.
Andrew Keen
Well, what about the idea Does this stop nowhere? I mean, you've got your AI book, Who Owns Intelligence. You told me, I think, that it could take you a few hours to write this thing. How are you supposed to compare it with someone who takes months, if not years, to write a book? Are we just...
Keith Teare
Well, so I would say that... Let's use this as an example. Unfortunately, you can't click on the read icon because it's an image. But if you clicked on the read icon, There's 15 chapters and each chapter has a narrative, a sequence of arguments with references. All of that was put together by me in about an hour and a half based on years and years of thinking where I'm aware of all those issues. I'm aware of all the writers in each chapter. I'm aware of the historical precedence. And so the structure The chapters, the arguments within the chapters was all prompted by me. What the AI did was to take that and produce a manuscript from it that then I could critique and shape, if you will, shape according to my will. Just like a color grader using Photoshop takes a picture and they use sliders to change contrast and exposure. They put light in places that it wasn't really there in the original. They shaped the photograph. And so I think the human part is 99%. The AI part is maybe 1%, even if 100% of the manuscript was dumped onto a page by AI.
Andrew Keen
This was also a week where Universities drop their AI detection tools over fizz, over their accuracy. Are you saying then as paranoia in the university, just as there is within the publishing business, that AI can be used to quote unquote cheat? Is it possible to cheat with AI or is in your mind any use of AI is legitimate and the C word cheating is redundant?
Keith Teare
Well, I think cheating is entirely appropriate against a set of rules that are no longer relevant. So if you take the rules as they are today, I do think people cheat when they use AI. I think the rules are at fault, that teaching is going to have to change. And the rules are going to go from did you use AI to how well did you use AI? And there's going to be differentiation between students based on quality, just as there always will. And somebody who's good at using AI will produce much, much better work than somebody who isn't. Because ultimately, the AI doesn't really do much of it. So I think there is cheating, literally. But to want to call someone a cheat, you have to be prepared to defend the old rules prior to AI and not think they should change.
Andrew Keen
And are you saying then that the rules established, say, for example, for the publishing industry, that they need to change?
Keith Teare
Absolutely need to change. I don't even know if they're rules, to be honest, Andrew. I think they're unwritten.
Andrew Keen
Well, they're unwritten understandings. The publishing industry is a very traditional old industry.
Keith Teare
Yeah. So there's kind of a snobbery around the art that Damien Hirst, for example, has tried to make fun of. The snobbery around art is a very narrow definition of what creativity is. And, you know, if you're going to have a dinner party, you wouldn't want an author that used AI to be there. Because, unfortunately, they don't...
Andrew Keen
Well, you're not going to get invited, then, Keith, to many dinner parties. Although, I'm guessing in Palo Alto, you wear your AI hat, so to speak, on your sleeve, and you wouldn't get thrown out.
Keith Teare
I think you'd still be disrespected. I think we're still living at a time when the assumption is, and it is a kind of a snobbish dated assumption, but it's real. The assumption is that to the extent you use AI, you're not legitimate, authentic, and so on. I don't think that's right, but I do think it's the dominant view.
Andrew Keen
So what becomes then of The artisan writer in the future, the Margaret Atwood, for example, who clearly doesn't use AI, who's hostile to it. What becomes of the people who have won the Booker Prize? Do they all just go away?
Keith Teare
There's still going to be the best authors who write the best stories who everybody wants to read. And whether they use AI or not won't even be a question. It'll just be, was this a good novel? Did it make a good movie?
Andrew Keen
Do you think it makes any difference in terms of the consumer? Today, when you go to the supermarket, and there's some controversy over this, we choose products because they're organic or natural. Will in the future there'll be a table in a bookstore, for example, that will say organic books written without technology, without AI? Is that one way that consumers will be able to... Articulate their own interests or preferences. I mean, you clearly don't care, but some people do.
Keith Teare
I think if they care about that more than they care about the quality of the book or the story, you know, I don't know what really motivates them. I don't know why they would care. But let's say, take your question, look, I'll go to the British Museum to see stone tablets with writing on done by Egyptians because it's fascinating, but it doesn't mean that that's the only legitimate form of publishing. And so everything is still legitimate and it may well be the case that an artisanal novel is fantastic. There'd be no reason to be against it because it's artisanal, but there'd be no reason to be against it If AI contributed to it either, it really is just about the quality of the thinking and the quality of the storytelling.
Andrew Keen
But there is, I mean, we can see it with music. People, when CDs came along, marketers boasted about perfect sound forever. There was a great debate. For a while, it killed vinyl. And now vinyl outsells CDs. And there's a lot more money to be made in vinyl than there are in... Digital downloads, it would seem. So maybe the future could be for entrepreneurs, at least. I mean, you're clearly not someone who's interested in this, but there could be a future for, shall we say, artisanal literature.
Keith Teare
There will be. It's said, see, for books, there definitely will be. And look, vinyl's a great example. Vinyl, these days, is manufactured from a digital master. So the vinyl is not analog. What it's doing is playing analog version of a digital master.
Andrew Keen
Well, you're not a vinyl guy, clearly.
Keith Teare
I've got a whole set of vinyl up here, but when I listen to it, it has crackles, and it definitely is not better than streaming, for example, from Tidal, which has 129... 192 kilohertz streams that are perfect.
Andrew Keen
Well, clearly, Keith, you're not sympathetic to the...
Keith Teare
I like live music, Andrew, where it's really analogue.
Andrew Keen
So you, I think that in the future though, I think that what you call artisanal literature or art or music, I think will have more and more, for better or worse, more and more value. Maybe it won't be better, but it will be different or at least perceived to be different.
Keith Teare
I don't disagree with you, but that won't be the market. That will be a little bit like billionaires buy paintings for millions of dollars. There'll always be scarcity and rareness leading to value that not everyone can buy into. And there'll be always a kind of a cultural elite that believes they're better than everybody else. And one of the signs of being better is that they don't use AI, may well survive in those elites. But it isn't a legitimate future.
Andrew Keen
Well, it's legitimate. If there's a business in it, I mean, just as you might have argued 30 years ago that vinyl wasn't a legitimate future, but it clearly is. I mean, if people are willing to buy stuff and determine their preferences on that, then that's legitimate, isn't it?
Keith Teare
It is, but it's tiny. I mean, vinyl compared to just one streaming service is tiny.
Andrew Keen
And then what about the university? It seems not coincidental that this crisis of AI has landed in the middle of a broader kind of pedagogical crisis within the university, of their place in the world, their cost, the role of academics and all the rest of it.
Keith Teare
Yeah. It's only going to get worse, isn't it? Because as we often hear from Sam Altman and Elon Musk and Dario Amidai, AI today is the worst it's ever going to be. In other words, it's going to get better and it's already gotten better over the last, you know, AI slop is a word you don't hear as much as you used to because there's less of it. It still appears, but it's less and less frequent.
Andrew Keen
So what do you make then? Do you disagree with Snapchat and YouTube and LinkedIn and Substack all trying to combat fake writing? Are they the King Canutes of this world? Are they struggling against something that is inevitable and that they're unwise to be doing it?
Keith Teare
I think they're about making money. And if the AI is so bad, That it puts consumers off from scrolling endlessly through their reels. They'll take it down. But it isn't because they have a quality threshold. It's because they have an engagement threshold. And I do think that especially in video and images, the quality of AI is still pretty low. It's kind of where AI was Two years ago, it's better than that.
Andrew Keen
Well, you use it on your editorials, although I always tease you, I always think they're dreadful, but maybe that's my elitist perspective. What about this word, Keith? Slop, AI slop, is that another meaningless word? Is slop redundant?
Keith Teare
No, it's not redundant. I think it's an accurate description of bad use of AI. And it's not unlike when Adobe PageMaker first came out and they had 20 fonts and really bad desktop publishing users put way too many fonts on a single page with lots of different colors and it was very ugly. People, unfortunately, who first use tools don't always use them very well. And so there is AI slot for sure.
Andrew Keen
I'm not convincing you very much, but I wonder whether you object to purely AI authored books. The interview of the week this week is with Andy Hunter, who's the CEO, a very interesting startup. We'll talk about them later in the show, bookshop.org. And he was interviewed this week for a New York Times piece about how AI books sneak their way into stores. And he, and I'm quoting him here, he said, are you going to allow scammers to rip off your customers? The answer has to be no, says Hunter. So they're banning AI written books on bookshop.org. Do you agree with that?
Keith Teare
Well, firstly, I don't think there's any such thing as an AI written book. It doesn't exist. There's books written by humans that use AI.
Andrew Keen
Right. So there are humans and he claims that some humans are creating these books in a few minutes and selling them as quote unquote legitimate books on bookshop.org and on Amazon. Is that okay? How are readers supposed to know?
Keith Teare
Why would readers care? Is the book any good is the question. If it's a bad book, then yeah, the readers should know. But they should know it's a bad book. There should be reviews of it.
Andrew Keen
So he's wrong that he's asked distributors like Ingram to filter out AI generated works. You think that's a wrong thing?
Keith Teare
I think it's outrageous. I mean, why would an AI generated book need to be filtered out? Can you even think of a reason?
Andrew Keen
Well, there have been examples in the past, for example, I could write a biography. I don't think there are any biographies of Keith Teare, but I could probably put one together using AI in about 20 minutes and then sell it on Amazon.com. Would you have a problem with that?
Keith Teare
You wouldn't get very many buyers would be the biggest problem.
Andrew Keen
Well, it might. I thought you were very famous, Keith. I've made you famous.
Keith Teare
But I don't think it would be wrong. Why would that be wrong? As long as it's accurate and interesting. Why would it be wrong? You're not really telling me why it'd be wrong. You're talking as if it's self-evidently wrong, but why would it be wrong?
Andrew Keen
Well, I think Andy Hunter would think it was wrong. I mean, let's say there's a biography of a better known person and someone put months or years of research and work into it. And then someone used AI to come up with a fake biography, but one that can be done in a few minutes. They're both side by side on Amazon.com. I think from their point of view as the gatekeepers, I guess, to book sales, they want to make sure that readers are not disappointed.
Keith Teare
Look, let's take the example of Winston Churchill. There's some very good biographies of Winston Churchill. I've read two or three of them. If we gave AI all of those biographies and then gave it the history of the First and Second World Wars and maybe the Boer War and the history of the British Parliament during those years and said, write me a book that positions Churchill As four different people in each of these four eras, influenced mainly by self-preservation, start with the Boer War and end with when he resigned from Parliament. And it produced a book that had all of those sources but was new, derived from my prompt. What's wrong with that?
Andrew Keen
Yeah, in a way, I mean, it's interesting that it's sort of becoming like Wikipedia or something. And people, if you want to know about the life of Winston Churchill and you don't want to buy a book, you might go to Wikipedia. But you're suggesting that AI will be able to replicate that. That's exactly what Gemini is essentially doing. I mean, now when you look somebody up, whether it's Winston Churchill or Keith Teare on Google, you come up with an AI, some people would call it slop, some people would call it More intelligent. But you come up with that story. But again, the question is, what becomes of the professional author? You're saying that the professional authors are the ones who cleverly put AI together to write this stuff?
Keith Teare
Certainly the best professional authors will figure out how to integrate AI into their work, for sure. And the ones who don't, because they choose not to, won't become worse. They'll just be who they are. But your description of having to spend a lot more time on a product that may not be as good, that will be what they buy into. And there's nothing wrong with that. That's their life choice. But it won't, their book won't be better because it took longer.
Andrew Keen
Yeah, it's rather like when, I don't know, when the engine was invented, or perhaps when we started riding horses, Difference between going from London to Manchester by horse or by railway or walking. So, yeah, so you're not, I mean, you're quite hard-headed about this. You don't have any sense of loss. I don't have a sense of loss. I mean, you're not really a book kind of guy. I mean, when was the last book you read?
Keith Teare
When was the last book you read? I'm just going to turn my camera.
Andrew Keen
You're ignoring my question. When was the last book
Keith Teare
you read? The last book I read was a science fiction novel by Robert Heimline called For Us The Living, and it was probably a year and a half ago. So you're not really a reader? I mean, you're one of these people who might read a book every couple of years. No, I don't read books very often. So you're not a book reader, so I'm not sure you're the right person to talk to about this.
Andrew Keen
Do you think you'll read more books if they were AI produced? Do you not read because you haven't got the time? You can't be bothered? Do you think they're a waste of time or you can't afford books?
Keith Teare
You're being facetious now, Andrew.
Andrew Keen
But I'm surprised that... You only read a book every couple of years and you have all these strong feelings
Keith Teare
about publishing. I think what happens in any discipline is over the years, this is a good insight into the meaning of the word intelligence. Over the years, you digest through reading lots and lots of things. And in my case, it's a mountain of books. You get to the point where you've developed a worldview. From everything you've read, your own independent view. And at that moment, real-time reading, newspapers, magazines, podcasts, blog posts, become much more important. A book takes a couple of years from concept to publishing. An article could have been conceived yesterday and be read today. So if you really want to be current, books is not The best starting point. I think if you're young and you haven't really yet formed a worldview, books are super important.
Andrew Keen
What about novels?
Keith Teare
I do read novels, but not as much. I mainly read nonfiction.
Andrew Keen
Well, you said you only read one book in the last two years, and that was science fiction. When was the last nonfiction you read?
Keith Teare
So you're doing it again. I actually think if you're mainly reading books, you're cutting yourself off from the flow of ideas.
Andrew Keen
Well, that's one position, Keith. So, I mean, really, so the destruction of the publishing industry wouldn't be a bad thing for anyone because it doesn't help anyone. So it may be authors who...
Keith Teare
Well, the publishing industry is basically a scam itself. I don't know, the book I just showed you sold about 50,000 copies. I think I got paid... A grand total of about ยฃ10,000. And the price on the cover was, what was it?
Andrew Keen
About 5p.
Keith Teare
You got about 5p a copy then. Yeah, it was ยฃ3.99. So, you know, who made the most was Penguin. They call something else out there, Penguin something else. And so the publishing industry is a middleman industry by nature. I think modern ideas don't go through intermediaries as much. They go direct from you to your readers, like our show here does. And so I do think books are super important. I want to read Plato's Republic. I want to read The Odyssey. But... Well, those are old books. Right. But well, they're important old books. They're foundational, as we say. But what I don't want is to only read books. And I think as you mature in your evolution of your thinking, books play less and less of a role and real time stuff plays more and more of a role.
Andrew Keen
I mean, it's clearly true for you. I'm not sure if that's true. It's certainly not true for me. It may not be true for everyone. Let's talk about the book publishing industry. As I said, my Interview of the Week was with Andy Hunter. He's doing something interesting. Keith, you've often talked about in our conversations about the redistribution of wealth in the digital ecosystem. I'm not sure if you had a chance to look at my interview. I did, yeah. Bookstore.org. He's doing something Quite interesting in terms of selling books to readers and then redistributing some of the revenue or much of the revenue to local bookstores. Is there something in the bookstore.org model, even if you're not much of a reader yourself, or a book buyer, a book reader?
Keith Teare
I listened to him and I liked him a lot. I do think that the victim of the publishing industry is not bookstores, it's authors. And he of course can't really influence that because he's taking books from firms like Penguin and selling them. And he can give a bit more to a bookstore for sure. But what you really need to do is to have direct publishing from an author to the readers, where most of the money can go back to the author. That's what doesn't happen. And I don't think bookstore.org solves that problem, which is a much bigger problem. I don't think anyone really cares about bookstores, to be honest anymore. If you can get a book tomorrow, even today, because you want to read it and on Kindle, you know, two minutes from now, then I think that serves consumers very, very well. It doesn't serve authors. So I think we should focus on solving how authors get paid, not how bookstores survive.
Andrew Keen
And then when it comes to authors, one of the platforms that claims at least to be acting in their interest is Substack. And there's an interesting piece you linked to this week on Substack and AI. They have a piece from Taylor Lawrence, a popular tech writer. What's Substack doing in our AI age to perhaps to help authors not just promote their work, but even earn a living?
Keith Teare
Well, it hasn't integrated AI at all into Substack. You would think that it would have integrated Claude or OpenAI as a tool for people to use, even if they only used it for correcting spelling. There's absolutely no software tools built into Substack. Until they partnered two weeks ago with an AI detection tool, which is where my title... You don't like the detection tool, do you? Well, it isn't very good, Andrew. I gave it my editorial last week, which was written by me, but from an AI draft. And it drew the conclusion that it was 100% AI. And I'd say a fair conclusion would have been probably 40%. So they're not very good at actually being accurate about the role of AI. And then the second thing is the assumption is that if it says that it's AI, that there's something bad about that. So the very act of doing this partnership reinforces this belief that there's something wrong with AI.
Andrew Keen
Yeah, I'm not entirely sure. So you used to be quite... Supportive of Substack. You're disappointed with them? Do you think they've joined the reactionary camp here with their AI detection tool?
Keith Teare
Look, they think they're being pro-author. They have a very clear worldview, which on the whole is great. Authors should be able to make a living from doing their authoring. I think that's a very good starting point. Now what they're starting to do is distinguish between authors who use AI and authors who don't. I don't know if that's really part of their Remit to do that. I think that's a mistake actually for their business.
Andrew Keen
So it's none of their business. I mean, I'm going to say in defense of Substack, but I use AI for some of my, certainly my postings about my shows. And it's quite easy to use Claude and then put it on Substack. So I'm not sure Substack necessarily needs to integrate it into their tool, do they?
Keith Teare
Yeah, I can agree with that. I just think it's interesting that they integrate an anti-AI tool, but no AI tools.
Andrew Keen
But there is a growing, I mean, I know you don't necessarily sympathize with these people, but there is a growing body of public opinion, which is hostile and suspicious about AI. Do you think these people will eventually come around, they'll just die off, or they'll read your stuff and realize that everything's AI, so it doesn't really matter?
Keith Teare
Look, I think everything will always be human and AI will always be useful as a tool for humans. And eventually that will be ubiquitous and no one will even comment on it any more than you would comment on, you know, the words of Socrates ending up on a 70 millimeter film on the screen. It'll just become normal that you get this separation Between a person and the form that they manifest their ideas in. And that separation used to be zero. You could only know my ideas if you were there and I spoke them. Then it became separate because I could write them down. Then it became separate because the book could be printed and go all over the world without me. And then it became separate when it could be turned into, you know, a radio show or a movie or a podcast. This separation that leads to content having its own life, separate from the creator of the content, AI is just the latest version of that.
Andrew Keen
The Economist, which you linked to this week, has a piece about how to spot AI writing. In another interview I did this week with a psychologist, she talked about the way in which people, particularly young people, are using AI to to figure out what to say to other people at dinner parties. They go to the bathroom, they take their phones with them, and then they plan their conversation. Is that okay, Keith? Is there any limits to the way in which we use this technology? Do you ever go to the bathroom during dinner parties and try to figure out what you should say using Claude or OpenAI?
Keith Teare
Never, because I'm not lost for words. But I wouldn't mind if an insecure person who was worried about what to say. I wouldn't mind if they got advice from AI. That would be okay.
Andrew Keen
No, we will see.
Keith Teare
I mean, we shouldn't make fun of people who are insecure about their presence, should we?
Andrew Keen
I would never make fun of anyone, especially the young, anxious people. I feel terribly sorry for all of them. What else is going on? So enough AI and cheating. You've made your point. I don't think I... I mean, I take your point. I'm not in strong disagreement, mostly, although I think when it comes to these AI-authored books, which are essentially scams on places like Amazon, I disagree. But anyway, what else is going on in Silicon Valley? Is the bubble continuing to burst? One of the big other stories this week. Is of a German character, a young man called Leopold Aschenbrenner. He looks like someone who might have been invented by Thomas Mann, whose $45 billion fund seems to have evaporated. Is the wind coming out of the boom, Keith? Or is this just a story of a curiosity? Some young kid who had a bit of luck and now has lost all that luck.
Keith Teare
Yeah, I think he's probably still super rich, to be honest, Andrew. I think this calamity is not necessarily a personal calamity for him. But look, this is the guy who was at OpenAI as a very young, I don't know if he was an intern or just a junior employee, and he got fired. And when he got fired, we don't really know fully why he got fired. He wrote a A long essay, which was a futuristic piece about what the world would look like once AI agents are fully matured. And there was a lot in that essay that prefigured what was happening inside the AI labs at the time.
Andrew Keen
Yeah, we covered it. And Tim Ferriss, author of The Four-Day Workweek, I think, Successful author called him the Nostradamus of AI, whatever that means.
Keith Teare
So then what happened? Well, look, he was young, attractive and articulate. Money started to pile into him as if he was going to be the best investor. And sure enough, it turned out that he could invest money and make money for his investors because he was riding the wave of AI. Given that these companies are doing a new funding round at a higher valuation more than once a year, it was inevitably he would make money. And he had the structure of a hedge fund. And it got to be worth $45 billion in a very short period of time. I mean, he didn't. His fund was worth it. His fund was worth $45 billion. It included lots of public market positions, and it included leverage, which is where you can invest N times more money than you've actually got. And suddenly, very, very quickly, he lost $600 million. And his leverage margin was called, and he didn't have the cash, so he had to get liquid. And he sold the whole thing to Citadel, another hedge fund. We don't really know the facts of that, but I'm guessing he was still in profit, and he's got a lot of cash in his hands now.
Andrew Keen
Yeah, this is Ken Griffin's What's interesting is that most of the value he had was in Anthropik. He still has that. Didn't he sell that? Isn't that the whole point?
Keith Teare
I don't think he did. So what is he selling? He sold the public assets, I believe, and maybe some of the private ones.
Andrew Keen
How do you sell public? Was he selling public assets at a knockdown price?
Keith Teare
You can't do that. You sell them at whatever price they're trading at.
Andrew Keen
So you just cash them in on the markets. What does Citadel?
Keith Teare
So Citadel took over his private positions as well. But not all of them, I suspect. And we don't really know yet. So let's just say that out loud. But he liquidated situational awareness, which was the fund. But I don't think all the positions have been liquidated. I'm pretty sure he's still sitting on some personally.
Andrew Keen
You're not, Keith, a great friend of Dario Amadai at Anthropic. Was this a week where Anthropic's public reputation nosedive finally? I'm sure you weren't particularly disappointed with some of the bad press it's getting. OpenAI hasn't had a great week either. But what's been happening in terms of how Silicon Valley in particular, other companies see Anthropic and their hostility to open source. Some of the articles you linked to are about open source, including a high profile piece by Mark Zuckerberg, of all people, in support of open source AI.
Keith Teare
Yeah. Look, the valley is full of sharks of different colors right now. Everybody's playing their own book. Amadai certainly has always played his own book. It's the only thing he knows how to do really. And you can't criticize him for that. Zuckerberg is zigzagging left and right trying to find a role for meta in this AI world. And his latest one is to support open source. And if you can believe it, the distribution of wealth to the people.
Andrew Keen
Shall we say a bit rich, so to speak, coming from Mark Zuckerberg. Coming from him.
Keith Teare
And Amadai refused to sign an open letter in support of open source, which all the other people did. And then he wrote an essay himself, which had the following sentence. I'm not against open source as long as it's safe. Now, he's on the record as saying that all AI is not safe. So to say I'm not against open source as long as it's safe is another way of him to say I'm against open source.
Andrew Keen
You must be enjoying his humiliation, or if not humiliation, certainly he isn't crying as much as he has been.
Keith Teare
I don't enjoy it. I made the point a few times to you, Andrew, that In a spirit of advising him, he's making a lot of mistakes. And I have a lot of time for Anthropic, and I use it a lot. But as a leader, he's making a bunch of mistakes. And most of his mistakes are either asking for regulation, that's one whole set, The second is not allowing the government to decide as a customer what to do with the tech it buys and thinking he can take over for the government as the policy maker. That was a mistake. And I think it's a mistake for him. To be honest, I think there's a very reasonable argument against open source, which he could easily adopt. And I wouldn't knock him for that. But he's being deceitful. He's saying he's for open source as long as it's safe. So he's not straight talking. So the problem with Dario is you can't trust him.
Andrew Keen
He's slippery like Sam. What about when it comes to open source in China? You have a piece on your LinkedIn from China talk about Chinese illustrators are becoming what the article calls tokenizable. The threat to, shall we say, the closed source or the The bounded platforms, the walled gardens of anthropic and open AI, ironically, is coming from China. But what does it mean for Chinese companies to be open source, given the nature of Chinese politics and capitalism?
Keith Teare
Well, strangely, China is the capital of open source right now. This particular story is is more aligned to our private prior conversation about authenticity, because it's about painters having to paint live on film to prove that they actually painted a thing. So that's more about authenticity. But China is the capital of open source. And a lot of the reason for that is that it's in a weak position vis-a-vis the US from a hardware point of view to do foundation models. It still has good foundation models, but by being open source, it gets to compete on the wall stage and it's working. So, you know, I think it's again, it comes back to self-interest or playing your own book. The Chinese playbook should primarily undermine OpenAI and Anthropics ability to make revenue from foundation models. And it also should make the world less dependent on those foundation models. That would be in the spirit of what would be a good thing for China. So I think it's doing exactly what it should do. And it's doing it so well that American corporations, the whole narrative has shifted to what's called routing prompts to AIs based on cost and quality. And the assumption is that a lot more queries will be routed to open source models at cheaper rates.
Andrew Keen
Yeah, I think my problem with the quote unquote open source is is defining what exactly it means. I mean, companies should perhaps, rather than talking about this open versus closed source, I mean, by definition, open source sounds better, but I'm always suspicious of things that by definition sound better. They really should simply be describing their business models. Companies like Anthropic and OpenAI have different business models from some of these Chinese AI startups.
Keith Teare
Yeah. Well, I think if you distinguish between consumer and enterprise, that's true. Consumers, you tend to install an app and you live through the app into AI. And so Codex or Anthropics Claude become very important. And owning the user interface becomes the key. And those apps are becoming everything apps. What does Elon Musk call them? There's a word that he uses like a Like the only app you ever need.
Andrew Keen
Like X, I mean, like he wanted.
Keith Teare
Like he thinks X is going to be. It's a little bit different in the enterprise because the enterprise you install AI on servers or you talk through the cloud through APIs and you're building your own user interface for your company data and goals. So there you can slot in an open source AI And it uses the same API calls as OpenAI does or Anthropic does. They all have the same APIs. So the user doesn't really know what they're using because it looks the same.
Andrew Keen
Yeah, I would advise listeners and viewers when they hear the term open source to... To get more clarification on what exactly it means. Finally, another thing that sounds sometimes almost too good to be true is Elon Musk's prediction that by 2036, Money Won't Matter, again, seems like, just like Mark Zuckerberg talks about open source, a bit rich coming from the world's first trillionaire, Elon Musk. But your post of the week is from another old friend of the show, Vinod Kostler, who had an ex-post suggesting that... This is not entirely unrealistic, I think personally it is, if the politics permits this great abundance. So the fact that Kostler is acknowledging that Musk might have a point, I'm guessing, Keith, you rather like this.
Keith Teare
Yeah, I wouldn't be as bold as to give a date like 2036. I think when you study economics, you learn this word, there's two words you learn. One is tendency. Which means there is a trend towards. And the second is keteris paribus, all other things being equal. And so all other things being equal, there is a tendency towards money becoming unimportant. But the difference between that and the money actually becoming unimportant, as Vinod points out, is a whole ton of variables.
Andrew Keen
Yeah, and he chose 2036 because obviously we're in 2026. So we're talking about another decay, another 10 years. I mean, when you think back to 2016, doesn't seem that long ago, beginning of the first Donald Trump presidency, it's hard to imagine. I mean, I know you like the idea of it, Keith, but it's really hard to imagine that money won't matter. It matters so much today.
Keith Teare
Well, you have to have a deep dive abstract conversation that also involves classical economics.
Andrew Keen
Well, you have to be as well read as you, even if you haven't read a book in two years.
Keith Teare
You do. I mean, the framework that you can understand it through is the framework of supply and demand. Money is a mediator between supply and demand. It is also... A representative of value. So it's a means of exchange and it's a store of value. Now, if there is abundance in things, a means of exchange does become irrelevant. And if the value shrinks due to the lack of labor time in it, value in the classical economic sense, It becomes cheaper tending towards zero in terms of value. So if you get an abundance of things which tend towards zero, it isn't illogical what they're talking about.
Andrew Keen
I'm willing to bet you.
Keith Teare
Anything within 10 years.
Andrew Keen
The whole point is you can't bet money. We could bet wives or children. Do you want to put your children and wives on the table, Keith?
Keith Teare
I don't know whether we'd be winning or losing with wives and children. We'd be losing, Andrew. We'd definitely be losing. Don't even go to there. Look, I think it'll be... By the time we get there, if it turns out to be true, the winner will have to And I'll talk to you next week. And by then, I assume that...
Andrew Keen
I assume that your Who Owns Intelligence self-published AI book is going to become a red-hot book. Is that fair?
Keith Teare
Look, I think if you're listening to us and you are an agent or a publisher, you should be talking to me because this book is putting down a marker for how the wealth from intelligence ends up benefiting all of us.
Andrew Keen
Well, there you have it. Keith Teare, not only the publisher of That Was The Week, but the author of the most important economics book, perhaps written in the last two or three hundred years, Who Owns Intelligence? I will see you next week, Keith. Bye.