Transcript Viewer
Sep 25, 2026 ยท 2026 #36. 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 2
Morning. Plan my trip. Looks good. Confirm. Trip confirmed. Schedule updated. Work's all done. Perfect. Thank you. Schedule updated. Great. Let's join the others.
Andrew Keen
Hello, everybody. It is Friday, the 25th of September, 2026, the last Friday of the month, it's been another astonishing week in technology. This time, the soap opera, if that's the right word, of AI and regulation and big and little tech in the United States, Silicon Valley, China. All moved to New York, to the United Nations, where AI leaders, quote unquote, at least according to some European media, warned the UN Security Council of risk to humanity as a whole. I'm not entirely sure what humanity as a whole means, but certainly Sam and Dario came to the UN to warn The United Nations of a possible peril absent a global response. Very interesting week. Lots of other European leaders, including Andy Burnham, Got involved, Australian leaders as well, getting involved over rogue open AI. And of course, Donald Trump played his role, suggesting that AI should be renamed superintelligence. He's much less worried than either Sam or Dario or Andy Burnham or the Australian Prime Minister. So this whole thing's gone international. And who better to talk about the internationalization of AI Our current AI crisis or opportunity than Keith Teare, highly international fellow, English-born, living now in Silicon Valley. Keith, another remarkable week.
Keith Teare
Yeah, boringly so, right? Because the parameters of the remarkable are incredibly static. On the one side, you've got doomers. On the other side, the anti-doomers. And the debate, which I thought was resolved, Last week clearly hasn't been resolved. I thought Amadei and Altman were put back in their box in a way when they promised with Elon Musk to control their own software stacks. I thought that was it. No more discussion.
Andrew Keen
Just wishful thinking, Keith, I think, in terms of putting those two guys back in their box. Neither of them seem to be your flavor of the month.
Keith Teare
Well, it's not really about taste. It's about logic. And neither one of them is displaying any signs of being scientific. And given that this is science we're talking about, It's incredibly political what they're doing.
Andrew Keen
There's almost no science. Keith has given them the worst insult, Sam and Dario. They are not scientific. Let's step back a little bit. Last week, actually, You misled me in our audience. You told me all this was over. So as you know, you were wrong, not for the first or the last time, like myself. What's changed? I mean, did they use... One of the things that strikes me about the UN is it's so insignificant that anyone can say anything, and it doesn't actually make any difference.
Keith Teare
Well, this is the UN Security Council, which has, you know, not very many, but some teeth. So it's not the general assembly. So it's basically spies and people who consider security. So they would both consider this is the, globally speaking at least, the biggest possible forum they could appear in. And they're not wrong to, I mean, AI has globalized, but not in the way that they, It's globalized through the spread of open source models. And every single country on the planet, there are people, individuals and companies using AI, most of which originates in the United States, but some of it comes from China. So globalization is real. But that's a technical fact. What they are focused on is What they want to see is political globalization, almost like a super state view of AI coming from the UN Security Council, setting rules. It reminds me a little bit of when Facebook launched their crypto token called Libra and stated upfront that they'd only actually ship it if they could get all the governments of the world to agree. And you and I spoke about it at the time. I said, well, therefore it's going to die.
Andrew Keen
And it did, of course. That's the last time I think we even talked about Libra.
Keith Teare
Correct. And this is very similar. Wanting to get the world's governments to agree on anything is super hard. Getting them to come together to put in place some rules that AI companies can use to defer liability from themselves. You know, hey, we're following the rules. Of course, when there are no rules, There's 100% liability.
Andrew Keen
So your position on these guys, and you're certainly not alone here, and I think your argument is credible, that they're doing it to maintain their advantage, that Sam and Dario, who don't agree on very much and certainly don't like each other, the one thing they agree on is if they get a little bit of global regulation of some kind or another, it will protect their competitive advantage. Is that fair, Keith?
Keith Teare
I think there's a second element. That is fair, but there's a second element, which is the discussion over the last couple of weeks has moved to liability. If something bad happens, who gets sued? And the only rational answer to that is the companies that made it happen. And given that they're both going for IPOs, liability goes away if you follow rules set by government.
Andrew Keen
And of course, both companies have lots of lawyers to mull over these things. Meanwhile, let's, I mean, we'll come back to Sam and Daria, because they're the central figures in all this. But meanwhile, what do you make of... Andy Burnham's call, for example, to make Britain the honest broker here, somewhere between the United States and China and the EU. He suggests that he can replicate what Gordon Brown did after the global economic crisis of 2008. Is this Britain once again, post-imperial Britain searching for a role in the world?
Keith Teare
I think it's Andy Burnham searching for something. Look, I must admit my familiarity with British politics is pretty shallow. I read the British press and that's about it. So I'm somewhat familiar, but what's certainly true and doesn't require any reading of the press is that Britain on the world stage is relatively irrelevant. And Andy Burnham on the world stage is relatively irrelevant, although he's about to become the chairman of one of the UN bodies for, you know, different governments take over for different periods of time. The G20 is about to become the chairman of the G20. It's Britain's turn. So what he's doing is he's setting the agenda for his G20 chairmanship. And he's going to focus on regulation, which is weird that he would pick this because, you know, the EU is widely believed to be an overregulator and Britain was meant to be a beacon of freedom. Certainly in banking, for example, companies like Revolut have come out of the UK because of the lack of regulation. And so Britain was this not regulator, almost Trump-like actually for the last 20 years. And he said, suddenly Burnham is grasping regulation as a good thing. I don't think that's going to serve the UK well. I don't think it'll serve him very well.
Andrew Keen
Yeah. And meanwhile, his counterpart in Australia, Anthony Albanese, He made news on two fronts. Firstly, he made some speeches at the UN also calling for international cooperation. There's also a rogue open AI story. Apparently it infiltrated the Australian government. I don't want to get into that one again, Keith. Is that just, I'm guessing in your mind at least, just an example of
Andrew Keen
badly programmed AI rather than AI being able to think for itself?
Keith Teare
Yeah, probably directed by a human who asked it to do very specific things. The fact that it was able to listen to the human is the programming piece. That's what Jensen Huang calls an engineering problem. But the fact that the human tried to make it happen is a social
Andrew Keen
problem. But on the other hand, so I take your point on that. It's easy to poo-poo all these attempts, but the Australians have been more effective than others when it comes to tech. Their regulation or legislation or banning on kids' use of social media has had quite a big impact. So it is possible for middle powers like Australia or Britain to actually shape technology policy, isn't it?
Keith Teare
I don't think they shape policy in the sense that It's not really going to impact what the companies do. I do think they impact the zeitgeist of their fellow governments and that leads to discussions, conversations, papers, sometimes legislation that Australia is a trigger or a catalyst for. That's certainly true. You know, one wonders what the upside is for Australia. In being anti-tech, certainly Australians are not anti-tech. They're one of the fastest adopters of almost all tech. And they have some great tech companies in Australia, like Atomos and Blackmagic, who produce all kinds of video equipment. So it is a little bit strange that the Australian Prime Minister wants to take that stance.
Andrew Keen
Are you disappointed, Keith, in the Prime Minister? Have you told him?
Keith Teare
You know, I have a good friend, Evan Thornley, who is a Labour Party member in Australia and was a regional minister. He was formerly the CEO of LookSmart here in the US during the bubble era. And I will give him a call and see if he can make a difference.
Andrew Keen
Meanwhile, Donald Trump, I always like talking about him, but he has made some AI news. He apparently talked about renaming AI superintelligence. He's clearly against all forms of regulation. He made a very defiant speech. Anything interesting on that front with Trump, Keith, or is it just business as usual?
Keith Teare
Well, no, look, the word superintelligence is not Trump's word. It's been called... Superintelligence has been one of the milestones on the map of open AI and Anthropics since day one. And they now claim that they've achieved it, even possibly artificial general intelligence, which is kind of in the same ballpark. I think, you know, I don't like... I don't like praising Trump because I disagree with him on so much, but I will say that when it comes to this, the core belief, which is Reaganite in a way, that free enterprise is best left alone in its formative years to evolve as fast as it can to the end goals that it is able to achieve is roughly the right instinct.
Keith Teare
You know, it's kind of sad to me that that is not a universally understood point of view. Certainly the demand on the democratic side, even Ezra Klein weighing in now for closing down this experimentation period is getting louder and louder.
Andrew Keen
Well, we'll come to Ezra Klein later. I'm not sure I agree with you on him. But anyway, It's creating some unholy alliances, this whole AI issue. I mean, Axios reports that Trump allies, quote unquote, are opening a new front against The anthropic CEO, Dario, over AI doomerism. On the other hand, so the Trump people see a culture war opportunity when it comes to effective altruism. On the other hand, the Financial Times reports that the MAGA base is recoiling as Donald Trump goes all in on AI, a kind of joke, of course, on the All In podcast. Steve Bannon, of course, last week appeared with Bernie Sanders in an anti-AI event. How is this impacting, in your view, the MAGA movement? Is it a big deal or is it just one more thing that they have to deal with?
Keith Teare
I think it's a storm in a teacup, to be honest, pun intended.
Keith Teare
I don't think it's real.
Andrew Keen
The culture war against Dario or the base hostility to AI?
Keith Teare
I think the base hostility to AI is not real. I think it's manufactured and I think it will have almost zero impact on how people vote. I think inflation is going to have a much bigger impact. I think the various wars will have a much bigger impact. I don't think the average person is walking around making voting decisions based on your attitude to data centers.
Andrew Keen
Although the FT piece is an interesting one. It reports on a woman called Amy Kramer, who founded Women for Trump, very much a A stop the steal type who's come out very openly. So there are some people, but I take your point. Meanwhile, with all the froth, the storm in the teacup that you talk about, your newsletter this week leads with a couple of very interesting interviews. I've watched or listened to them both. Ezra Klein's interview of Jensen Huang, which you call an interview for the ages. And then I thought an equally interesting interview between two people who have appeared on my show before. Mick Thompson, the CEO of Atlantic and Azim Azar, the Golders Green based technologist.
Andrew Keen
The founder and CEO of The Exponential View. Let's deal with this Hwang Klein interview. I have to say, I thought it was interesting, but I disagree that I take your point on Hwang. I'm not sure Klein was quite as combative as you suggest. It wasn't so much a debate as an interview between a skeptical journalist and a very smart technologist. Did you see it really as a debate?
Keith Teare
No, no, I didn't. I agree with you on the tone. But Ezra Klein did his own story a few days earlier where he was pretty clear that he was on the side of the doomers. That did not really manifest
Andrew Keen
because he was a good- I haven't listened to, so I can't comment on that. But judging from, I mean, I listened to Klein quite a lot, and judging from his conversation with Huang, he didn't sound to me like a doomer, but that's another question.
Keith Teare
Well, you probably did notice things like, you know, when Huang said that he doesn't believe you can trust Altman and Amadeus' points of view as being valid, You see Klein kind of scum and going, really? You know, which is even as an interviewer, he couldn't really resist display.
Andrew Keen
I have to say, you're a little sensitive on that. I mean, I would have given the same expression. Doesn't make him a doomer. He can be skeptical.
Keith Teare
Well, the facts are out there about Klein. I mean, Klein has gone from his Abundance book to echoing some of the Bernie Sanders, Elizabeth Warren, We've got to be super careful about this point of view.
Andrew Keen
You don't have to be falling. I don't want to turn this into a debate about Klein. You can be skeptical of some of the AI stuff without falling into the Warren, Sanders, Bannon camp, can't you?
Keith Teare
I don't think you can be skeptical about the idea that AI is species-like in its ability to set goals and subvert authority. Right. Klein makes it clear that he doesn't believe any of that.
Andrew Keen
Anyway, let's move on to Huang. I found it particularly interesting from my point of view, and I sent you a note on this before we went live. I was really struck by... With how, and maybe I'm putting it in the wrong language, how he threw both Dario and Sam under the bus. He was very critical, openly critical of both Anthropic and OpenAI. I think he even mentioned them at times in terms of their, quote unquote, doomerism or their fear of AI. Maybe you can explain what he said and what you make, given that he's a major investor in both companies and they're enormous customers of his.
Keith Teare
Well, look, remember that he's trained as an engineer. All of his training is as an engineer. And engineers build things. And most of the time, they fail before they succeed. For example, the Golden Gate Bridge wasn't the first ever bridge to cross water. And most of the early attempts to do that didn't work. Or the first airplane that flew wasn't the first airplane that tried to fly. All engineering is about failure leading to success. So his demeanor isn't to deny challenges. He doesn't deny challenges at all. But he characterizes them as engineering problems, not as species-like behaviors that can't be controlled.
Andrew Keen
And his argument, and correct me if I'm wrong, his argument, which was very explicit and made a lot of headlines, It's an excellent interview. I would suggest everyone watches or reads it. It's quite long. It's about an hour and a half. His argument is that if Anthropic and OpenAI, as they suggested in their warnings to the United Nations this week, if their products offer a possible peril, then they simply shouldn't sell them. Is that a fair way of capturing, summarizing Huang's argument?
Keith Teare
Yeah, actually he used exactly the same words that you and I used on this show. I used, I don't know if you agree with me, a couple of weeks ago when I said they should be shut down. And you asked me, do I believe they should be shut down? And I made the point that if they're right, that it's out of control, absolutely, the whole thing should be shut down. And Jensen Wang literally said, the labs should be shut down. Now, exactly like me, he was being facetious, because he doesn't believe AI is out of control, therefore, he doesn't really think they should be shut down. But if you believe Altman and Amidai, the only conclusion you could draw is shut them down, not regulate them, because how, How is a regulator going to do more than they can do at controlling AI? Setting rules doesn't give you control. It just gives you a set of rules to break.
Andrew Keen
So what's your set? I mean, clearly this was set up. Klein flew out to California to do the interview. Huang doesn't do many interviews. It was chosen by him and his PR people. They were making a major statement.
Speaker 2
Are they calling...
Andrew Keen
Sam and Dario's bluff.
Keith Teare
What is he trying to do
Andrew Keen
here?
Keith Teare
He's trying to change the discussion and turn Sam and Dario into manipulative liars.
Keith Teare
Basically, he's accusing them of being manipulative liars. And I agree with him 100%. Yeah, but that wasn't my question.
Andrew Keen
I mean, this is not church. He has goals as the CEO of NVIDIA.
Keith Teare
His goal is to allow AI to continue to grow because he sells hardware that fuels it. And he probably believes that if Sam and Dario don't zip their mouths...
Keith Teare
An unintended consequence is that AI will be not regulated, but constrained to the extent that it can't grow anymore.
Andrew Keen
So he's telling them, and I'm sure he's done it privately, doesn't seem to be the kind of guy who's particularly shy. He's telling Dario and Sam that if they're genuine about what they say, they should shut their labs down or certainly not ship product. But as I said, is he calling their bluff? Does he know that this is just a way of putting political pressure on them to get them to shut up?
Keith Teare
Look, as an engineer, he knows that what they're saying isn't true. Why isn't it true? It isn't possible for it to be true because AI doesn't have Yeah,
Andrew Keen
we've been over that one before.
Keith Teare
No, but Andrew, this is a really important engineering point. It is not capable of making decisions.
Andrew Keen
But that wasn't my question. We've been over that
Keith Teare
one before. Well, that's his core.
Andrew Keen
But he's not an engineer. He's talking as the CEO of the most valuable company in the world, a company that's worth about more than $5 trillion. So... This is not some intellectual exercise. This is him articulating the interests of his company.
Keith Teare
Well, are you saying that he doesn't believe what he's saying? He's just saying it for commercial reasons?
Andrew Keen
No, I'm sure he believes what he's saying, but that's less interesting.
Keith Teare
And do you think that Sam and Dario believe what they're saying? Because I don't.
Andrew Keen
I certainly believe that, I think Jensen's more genuine than Sam and Dario. Right.
Keith Teare
So his engineering call leads him, I don't think he's pursuing commercial goals at all. I think he's angry as an engineer.
Andrew Keen
Yeah, I don't agree. I think you can't have it both ways. You can't have your cake and eat it. On the one hand, everyone has their own interests. And you always use that argument with Sam and Dario to criticize them. But when there's someone you like, they're speaking in universal humanistic terms.
Keith Teare
I've never said that I like Jensen Huang. I don't even think about it in those terms.
Andrew Keen
Well, you like his arguments.
Keith Teare
Well, that's a totally different thing.
Andrew Keen
Yeah, but it's... I'm not suggesting he's necessarily wrong, but there is...
Keith Teare
The thing that really intrigues... There's no subjective part of me assessing his arguments. It's just, does it make sense?
Andrew Keen
Yeah, and, you know, I get his arguments and I'm sure... I mean, I'm not an engineer, so it's impossible for me to comment on that side. The one area where I thought he was weaker was this idea of always knowing. If your products can do harm. I think if we had some of the, maybe not the doomers, but the skeptics on the show, they might say, well, it's still not clear. Daria Huang presents very much a black and white world. Either the products can be shipped and they're good or they're not and they shouldn't be shipped. But the reality, it seems to me, of AI is that nobody quite knows. It's still...
Keith Teare
Very, very unclear. I don't agree. Nobody knows. I mean, if you think about most of the stories that have come out have been unreleased models in the lab. I do believe that both Anthropic and OpenAI have responsibly shipped products that they are prepared to get behind. And they haven't shipped things that they don't have control of. But in the lab, by design, They try to find out what the parameters are. And by definition, that will allow things to happen that wouldn't happen out in a release product.
Andrew Keen
No, I take your point. But let's just use the social media example. I mean, hindsight is always 20-20, it goes without saying. But had we known some of the stuff, and you brought up Facebook earlier, and Facebook have an interesting AI product that came out this week that I want to talk about later in this conversation. But had we known the impact or some of the impact of social media on society, I know you skeptical on some of this, but you acknowledge other parts, then perhaps there would have been a better case for regulating it or maybe not killing it, but certainly regulating it when it came out or when Facebook came out in 2008, 2010. My point is that
Andrew Keen
It's all very well talking about engineering and products doing harm from a very black and white point of view, but it's still not. Leaving aside the apocalyptic side of AI, which I take your point on, it's It's quite hard to know what kind of impact this technology is having. And listening to Wang, I mean, he wasn't a utopian, but he certainly had a radically optimistic take on AI. He thinks it's going to benefit everyone, turn everyone into entrepreneurs, blah, blah, blah. So I think that's, in my view, at least the weakness of his argument, or at least his performance with Klein.
Keith Teare
Yeah, but I frame it differently. I think we are living at a time When we are turning human beings into victims and blaming tech for the victimhood, social media is a great example. Sadly, when a teenager commits suicide, it is often today 100% blamed on social media when in fact that teenager's parents, life, Friends all played probably much more major roles in the outcome than any social media. And so we're demonizing tech as causing human victimhood. And humans aren't victims. Humans are completely capable of making independent decisions and nothing tech does can change that. So the idea that we live in this very vulnerable world where we have to think about everyone and therefore police everything to protect people is the exact opposite of the truth.
Andrew Keen
Yeah, well, again, this is a broader subject. I mean, I take your point, obviously, and my wife, was the head of litigation at Google in these recent social media trials. So you have a point. But that doesn't mean that there are sometimes examples where tech is having a problematic impact on society, on individuals, and can't be regulated.
Keith Teare
I can tell you, Andrew, in the last week, I've spent way too much time looking at these things online, lenses. And social media has given me the ability to see every lens in the world for sale and scroll through them one by one and listen to reviews. And I can guarantee probably 20 hours of my time has been completely wasted. Now, should I blame social media for that? Or do I blame me for that?
Andrew Keen
No, you're not a teenager. Let's move on on this. Let's go back to the Klein-Hwang interview. The most interesting thing, because as you know, The thing hanging over this is this threat or opportunity or promise of regulation. Certainly Klein is more sympathetic to regulation than Huang. And I thought the most interesting interaction was that they were talking about the amount of investment NVIDIA has in The AI industry and Huang explained that they had over $100 billion of investment. And Klein kind of muttered under his breath, well, that's more than the Biden Chips Act put together. And it really, to me at least, spoke of the way in which the state doesn't have the central role. It's not really the big player anymore in the same way as the UN can Say whatever they want, and they can hold all these fancy forums, but they don't actually make any difference. I think what this conversation, which I agree with you, was a very good one, and I think both were excellent, speaks of the way in which companies like Nvidia, for better or worse, are actually driving the innovation economy.
Keith Teare
Yeah, well, look, they can attract capital Due to their revenues. NVIDIA is actually sitting on tens of billions of dollars of cash, a huge pile of cash, as is Google, as is Amazon, as is Microsoft. They have huge piles of cash due to the fact that people pay for their products. And governments are missing two things. Firstly, they're missing vision. So even if they had the cash, You don't trust that they do anything useful with it. And two, they don't have the cash. They're in debt, huge debt, due to the way the world economy works. So we are not living through an era of
Keith Teare
state-led innovation. We're living in an era of bottoms-up innovation. And Silicon Valley, weirdly, which is a dot on the world map, is where 75 to 85% of it is happening.
Andrew Keen
Yeah, I agree. I mean, but things have changed. I mean, there was a period in certainly in the 20th century where governments were much more centrally involved in innovation from the National Health Service to all sorts of other investments in industry. Some people are nostalgic for that age, but it's passed. And this leads us to your second interview, which was also Really good. A little bit more technical in an economic sense between Nick Thompson And Azim Azhar. And I thought what was particularly interesting was the way in which Azhar explained that NVIDIA isn't really that dependent on OpenAI and Anthropic for its business and that it works very well also with the open source Chinese companies. In other words, coming back to my question about why Hwang was so critical of OpenAI and Anthropic. It's that NVIDIA's success doesn't depend on them. And I didn't quite understand that, but certainly Azeem made that quite clear. It was a particularly interesting conversation. He's also cautiously optimistic we're not in a bubble. I think Nick Thompson is slightly less optimistic, but Azeem also recognizes that nobody quite knows.
Keith Teare
Well, he does give numbers as to why it's very unlikely we're in a bubble. And those numbers are to do with the ratio.
Andrew Keen
I would say unlikely. I wouldn't say very. I don't think he would say very, but putting words into his mouth.
Keith Teare
It's one word. I mean, you might be right. But look, the main point is that OpenAI and Anthropic are renters. What does that mean? They don't buy GPUs, they rent them. They rent them from Elon Musk, for example, in the case of Anthropic, and from Google in the case of Anthropic. They are mostly, historically, have not built data centers. I think there is some change there. I think OpenAI is committed to building data centers now, but hasn't actually got one yet. Whereas Jensen Huang's customers are buying GPUs to put into data centers or offices or whatever. OpenAI and Anthropic are not direct customers, but they are indirect customers because they're renters, and they rent from the people who buy from Jensen Huang. So it is an integrated economy where any purchase of a GPU is good for Jensen, and OpenAI and Anthropic are dependent on the availability of GPUs. So Azeem is both right but also indirectly wrong. Um, about, uh, the dependence and given the, if you measure, uh, the number of tokens, uh, Peter Walker used to be a Carter is now open router and on LinkedIn has started producing very, very good charts about which tokens are people paying for on open router, which is one of these large aggregators, um, of all the different models. And OpenAI's token purchase is way ahead of
Andrew Keen
everyone else's.
Keith Teare
Now, it is true that open source models are growing as a percentage, but they're still small compared to OpenAI. And OpenAI's price per token is much lower than Anthropic's price per token. So if you measure it in dollars, Anthropic's doing better. But if you measure it in tokens, OpenAI is doing better. Very good person to go and follow on LinkedIn, by the way.
Andrew Keen
But one of the things that struck me having listened to both these conversations is that the success of the AI revolution or the AI economy in terms of whether it's a bubble or a bus, whatever you want to call it, isn't entirely dependent on the success of open AI and anthropic. It's quite conceivable that the AI economy can prosper without open AI and anthropic.
Keith Teare
Yeah, if you think of it as the token economy, which kind of abstracts from the companies to what it is they sell, every single software interface on earth is going to be offering user experiences that require token consumption. And so the consumption of tokens is just going to get bigger and bigger and bigger and bigger. The number of GPUs you need to produce tokens is going to shrink. Because they'll get better and better at doing it.
Andrew Keen
And the price will come down as well.
Keith Teare
And the price will come down. And this week, not to segue unless you're ready to, but this week, Facebook literally made it free.
Andrew Keen
Right. I think if we're to generalize about this week, whatever Dario said or didn't say in the United Nations hasn't been a great week for either of them. Their major supplier and a major investor, the biggest player in the market, was openly critical of them. Meanwhile, as you know, Facebook came out with its own product, Muse, Is it a direct challenge?
Andrew Keen
You linked to a very good piece by M.G. Siegler, who writes a very compelling argument for why Facebook is back and why Mark Zuckerberg has found his groove again. Is Muse, Facebook's AI product, its agent product, is it direct competition to anthropic and open AI?
Keith Teare
Yeah, direct. It's a full stack. It has large language models at the base. It has various logic and business services in the middle. And it has a user interface, actually many user interfaces that people can use on the top. And it's an always on doing agent, an agent that does things for you.
Andrew Keen
I know you love playing around with technology, whether it's lenses or AI agents. I assume you've been giving it a run. Is it any good?
Keith Teare
I actually haven't used it, to be honest.
Andrew Keen
You can download it, right? Yeah,
Keith Teare
you can download it. U.S. only, by the way. So if you listen to this outside the U.S., it won't be available. But in the U.S., everyone can use it. So it does the same as Claude? No, it does the same as Open Claw and GrokBot.
Andrew Keen
But for most people, it's sufficient. I mean, my reading of this and in the conversations you and I have had, would it be fair to say that Muse is good enough for most people? It might not be quite good enough for you. You're a high-end user. But for most people, it's sufficient.
Keith Teare
More than sufficient. It's the next generation. What they've done...
Keith Teare
This is a little bit futuristic, but by futuristic, I mean the next six months, not the next six years. They've taken the computer that we all use. So to use Claude, typically you're using a computer or an iPhone. They've taken the computer and they've put it in the cloud as a virtual machine. And they've given you a visually distinct agent. You design the agent. It's look and feel. It's a kind of animation. Like a Toy Story character or something. And it controls the computer. And you talk to the agent with your voice. You can type to it as well. It speaks back to you through speakers or headphones. Or it can type back to you. And if you want to send an email, you tell it what to write. And it uses the computer to do it. So it's taken the need for a computer away and put it in the cloud and given you access to an intelligent agent can pretty much achieve any task as long as you give it the tools.
Andrew Keen
But you do need an iPhone. You do need a device to use it. It doesn't just pop into your head. You can't just use it on their glasses.
Keith Teare
Yeah, you use it on their glasses. You can use it just on earphones. There's a new set of glasses that has no cameras, only earphones. So you can use it there. It has earphones and a microphone. They produce this little old-fashioned watch-like thing that you put on a chain. That has the agent on. So you don't even need the glasses. You can just talk to it using that and use Bluetooth to connect your earphones to it. So it's totally abstract. The agent is the permanent thing. The device has become second level.
Andrew Keen
The hardware, you mean? So two questions on this. The obvious question, of course, Keith. I'm sure everybody's asking is if it's free, what's Facebook's business model? Their business model has always been controversial, especially with social media in terms of advertising. How are they going to make money out of this?
Keith Teare
So they announced deals with almost every single retail facing store on earth. Well, not on earth, in the US. You know, Target, Walmart, CVS, Instacart, everything. So you can basically have your agent shop for you. And their business model is to take a percentage of every transaction.
Andrew Keen
What happens if you don't buy stuff online? Then they just lose?
Keith Teare
You'll get it for free. Their model obviously says that across 3 billion people, enough people will shop for them to do very well.
Andrew Keen
So it doesn't come with advertising. I've noticed sometimes when I use OpenAI and I haven't Logged in that I get advertising with the interface.
Keith Teare
Yeah, it doesn't come with advertising. And even more striking for Facebook, they're adopting Apple's stance. They're creating encrypted stores for all your stuff, all your interactions, that even Facebook doesn't have the key to unencrypt.
Andrew Keen
And speaking of Apple, surely, and Microsoft also came out and you have a link with an interesting interview with their CEO, Nadella. They came out with a new co-pilot this week. You have to assume that Apple will come out or must be working on something relatively similar, maybe with a particular Apple slant on all this.
Keith Teare
Yeah, well, for Apple, this is a scary moment because and I should expect the same is true for Google. We are probably moving. You know, I made the title this week, What was it? The Future Just Showed Itself.
Andrew Keen
Not one of your best titles.
Keith Teare
But it's descriptive. What Facebook just showed us is a computerless intelligence capable of acting on your behalf, under your direction.
Andrew Keen
Well, when you say, I think it's a bit confusing, when you say computerless, you still need some sort of smart connected device. It might be headphones, it might be earphones, it might be glasses, but you've got to have something. It's not going to be tied into your brain.
Keith Teare
Let's not go there, but I suspect he'll...
Andrew Keen
Well, eventually maybe it will in a few years, but for the moment...
Keith Teare
But for now, for now, You know, I would differentiate a lot between an iPhone and a set of earphones. I think what Facebook have done is reduce the hardware you need down to something with Wi-Fi, a microphone, and the ability to connect to earphones via Bluetooth. That's all you need. Now, that is a much, much smaller, probably sub-$100 device to build compared to a $2,000 device for an iPhone. So you really are reducing the cost of connection to an intelligent agent down to probably $100.
Andrew Keen
Plus, I think what this does is it does away with the textual relationship which OpenAI and Anthropic are still dependent on and turns it presumably into audio. You talk to this agent on Muse, do you?
Keith Teare
You can. That's the dominant form of interaction is voice. But obviously there are many times when humans, it would be very inconvenient to have to use voice in all kinds of scenarios. So you can also type or you can write with a pen and use the camera to have it look at what you wrote with their most expensive set of glasses, which are Kind of like Apple's, what's
Keith Teare
it called, Vision Pro. But glasses, only glasses, they're going to cost $1,300. They're as good as the Apple thing, as far as I can see. Well, they project a keyboard on 20 flat surface, and they demonstrated typing, highly accurate fast typing, on a completely virtual keyboard. So you could be sat on an airplane and just pull down your food tray, project a keyboard down there and be typing on the food tray. Everyone around you would think you're typing on the food tray. But actually, there's a really, really good keyboard taking your inputs. So you don't have to use voice.
Andrew Keen
So you're excited by this, MG Seagler is. Again, can we conclude this week that it's been a bad week for... Sam and Dario, and certainly a good week for Mark Zuckerberg, and a good week for Jensen Huang. He's become even more powerful and influential after his Ezra Klein interview. Presumably, Jensen benefits from, or certainly NVIDIA benefits from this Facebook initiative. Is that fair?
Keith Teare
Yeah, look, I think it's a bad week for OpenAI and Anthropic from a... A public policy point of view. Their PR strategy is in tatters this week. From a technical point of view, I'm less certain because I think both of those companies have enough built to very quickly emulate what Facebook announced very quickly in weeks is my guess. They're close to it already. As you said, Claude Claude, you know, it still lives in your computer. And you need an expensive laptop to do a good job of using it. But that won't be true for much longer. Remember, Johnny Ive is open AI working on something we don't know why.
Andrew Keen
Yeah, we will say on that one. I'm not holding my breath with him. But we'll see.
Keith Teare
Yeah.
Andrew Keen
Well, another remarkable week. It will continue, I think, Keith, to remain remarkable. We will talk next week. Not a great week for OpenAI or Anthropic, certainly on the marketing point of view. And I strongly suggest everyone watches Klein's interview with Huang and also Thompson's interview with Azim Azhar. They're both excellent and Very educational and making sense of this complicated and very fast-moving economy. So I'll see you next week, Keith. Keep well. Bye, Andrew.
Speaker 2
Morning. Plan my trip. Looks good. Confirm. Trip confirmed. Schedule updated. Work's all done. Perfect. Thank you. Schedule updated. Great. Let's join the others.
Speaker 2
Thank you.