Speaker 1: Just go back a week ago, when you think about that picture of the tech bros next to the president and then them announcing half a trillion for Stargate, this big project for AI, and it just looked like American tech supremacy was set for decades.

Speaker 2: But that US dominance only lasted for four days because a new Chinese AI model is taking the world by storm and causing chaos on the financial markets as a result. We'll talk about that on this episode of Newscast, the BBC's daily news podcast. But first, we're going to dive into the quite unexpected bromance between Donald Trump and the UK Prime Minister Keir Starmer. Hello, it's Adam in the Newscast studio. And shortly, I'll be joined by two of the BBC's top experts on AI to talk about the Chinese artificial intelligence model, which has basically sent financial markets into a bit of a meltdown, especially for the stock prices of some big tech companies. So there's lots to unpick there. It's absolutely fascinating and a very kind of now story. Another very now story is how Trump's second term as president is shaping up and it's shaping up to be one where he's on quite good terms with somebody who's from a completely different political ideological persuasion from him, Keir Starmer, Prime Minister of the UK. First of all, we had these very warm words from the president on Air Force One when one of our colleagues asked him about Keir Starmer. He said that they got on pretty well, even though they didn't see the world in the same way politically. And then we had a 45 minute phone call on Sunday between the two men themselves, followed by a quite effusive readout, as it's known, from Dining Street about the contents of the phone call. So let's get our two best London and Washington whisperers on the line. In Westminster, it's Chris Mason. Hello, Chris. Hi. And hello, North America editor, Sarah Smith. Hi there. How have you warmed up since the inauguration, Sarah?

Speaker 3: It was one very cold day. The weather has improved. The snow's still on the ground, but yeah, the thaw is coming.

Speaker 2: Now talking of warmth, that's quite a good word to describe this phone call that Donald Trump and Keir Starmer had over the weekend, Chris.

Speaker 4: Yeah, I mean, it was kind of conventional, really. It felt like conventional diplomacy, at least in terms of the publicly shared kind of sentiments from each side in what are known as the readouts. It's conventional when two leaders speak that each side will kind of churn out a few paragraphs about what was said on the call. And the two things have to sort of broadly match, so it looks like they're talking about the same conversation. But the emphases are always interesting and the language is always interesting. But for me, the story of the weekend was one of the BBC asking President Trump about Keir Starmer, you know, on microphone on Air Force One and the remarks being positive. And then the readouts from the calls being, you know, pretty warm and pretty positive too. And I think certainly from Downing Street's perspective, I mean, this is not the greatest political insight I'm ever going to offer, but that's better than the alternative. And, you know, ahead of the likely trip that the prime minister will make to Washington, I suspect in the next two or three weeks, three or four weeks, roughly, that's a kind of useful basis upon which to start. You know, I think both sides are aware that there's significant differences of character, significant difference of political outlook, and there will be significant differences on particular policies, no doubt, to come. But as things stand, the exchanges are pretty warm. And as I say, pretty conventional, which is newsworthy, given that the new or returning incumbent in the White House is anything but usually.

Speaker 2: Because yes, Sarah, if Donald Trump doesn't like somebody, he says it.

Speaker 3: Oh, yeah, he's pretty clear about it. And he would probably have tweeted or put on X's for use about it already, too. And, you know, there's no reason to think it wasn't a warm call, although that's not a word that appeared in the readout I got at this end. So Chris is right. They're often broadly similar, but it's where the differences come that it's always worth looking at. So the one that the White House sent out said that they'd held a call where Trump had offered his condolences for the recent loss of the prime minister's brother, expressed his well wishes for the British royal family, which is the great Trump card, of course, that the UK holds. Donald Trump loves the royal family and being invited to royal palaces and things like that. So that is something that we've got to offer him. And then it said that the two leaders discussed the recent release of an Israeli-British hostage and also how both countries can promote a fair bilateral economic relationship. I think the word fair is doing a lot of heavy lifting there. But warm wasn't something that came into it, although he did say very encouraging things when he was asked by the BBC on Air Force One about Keir Starmer, saying that he doesn't always agree with him, but that he has a very good relationship with him and that they'd enjoyed meeting each other a couple of times, that they'd had dinner before.

Speaker 4: I was threatening Adam to someone high up in government at this end when I was asking for a bit of interpretation, a bit of extra detail on top of the what a Downing Street spokesman had said and sent out to journalists. And I'm just looking at my notes now and you'll notice a recurring word. Very warm, very good, very personal, very positive, very encouraging. Very, very. Exactly. Exactly. And but they also acknowledged it was lots of talk about working together rather than policy detail. In other words, agreeing not to talk about the stuff that they know they disagree about so that therefore you create a conversation where it is warm and it is pretty agreeable because you've not talked about the stuff where, you know, where there will be disagreements.

Speaker 2: And Sarah, we've got a convenient list from Pippa Creerer, the political editor at The Guardian who posted on social media all the things that it sounds like they didn't talk about where there might be disagreement, starting off with the identity of the UK's new ambassador to Washington, Lord Mandelson.

Speaker 3: Well, so there's a little bit of controversy around this because there was a story appeared in the British press saying that there's a possibility that Donald Trump will refuse Peter Mandelson's credentials when he arrives as the new UK ambassador. So that's a formal thing that you do when a new ambassador arrives. You have to present a letter or something to the administration and they sign off that, you know, they will now speak to you as the UK's representative. And I haven't been able to find out at this end whether there's any truth to that. I mean, it came from somebody in the Trump camp, but whether that was somebody just expressing their own musing or whether it really came from the top has been very, very difficult to bottom out. We haven't seen Peter Mandelson here yet. I thought it was kind of notable that he didn't come for the inauguration celebrations when there were lots of parties being held around town that our current ambassador was very present at. That would have been maybe an opportunity for him to introduce himself to some people in the upper echelons of Trump world, or maybe it would have been seen that he was stepping on the current ambassador's toes if he did turn up at those things. But yeah, all the Brits are waiting with bated breath to see whether or not he really will be accepted as the ambassador when he turns up probably next month.

Speaker 2: And then, Chris, one of the other things on Pippa Correa's list, and maybe we won't go through the whole list because this podcast will then be kind of Joe Rogan three hours long, but defence spending as a percentage of GDP, because Trump has talked about NATO countries spending 5% of GDP, which is actually more than even the US spends at the moment. Labour are sort of sounding a bit lukewarm on even spending 2.5%.

Speaker 4: Yeah. And so, yeah, that's, you know, that's another potential, if not flashpoint, then point of disagreement. Equally, we know that a key part of the President Trump playbook is to start a negotiation with a big number in the hope that you edge people who are miles away from that big number vaguely, as he would see it, in the right direction towards it, even if the likelihood of very many getting anywhere near it, including the States, is probably pretty limited. And I think actually on that particular issue, I think it's broadly speaking in sentiment terms where the White House is pushing against an acceptance from plenty of European capitals, not least here in London, that the argument that he made vociferously and bluntly in his first term around NATO and frankly other members not pulling their weight as he saw it, that there's been an acceptance from others that he's right about that. And actually, in the passage of time since, as the world has arguably got considerably more dangerous still, the nature of that argument and how it is more compelling to more people, I think is just more prominent. Now, that is separate, of course, from the specifics of at what point might the government here be willing to put a time commitment on its desire to get to 2.5% of national income spent on defence. Intriguingly, I was talking to one senior MP here earlier on who was asking an even bigger question really, which is that the conversation in recent years has always been about that proportion of GDP. But if you're in an era of relatively low growth, at what point do you get to a conversation where it's actually about what your defence forces need, as opposed to a arbitrary percentage of an arbitrary number, which may or may not add up to what armed forces are deemed to need in what looks like a kind of dangerous generation ahead.

Speaker 2: But Sarah, we've learned from Trump, the thing he tends to care about when it comes to defence spending is the number.

Speaker 3: Absolutely. And he does want to see a lot more money going in. He's made that very clear, not least because he points out that in terms of dealing with Russia and Ukraine, the biggest issue that NATO is facing at the moment, the US is geographically nowhere near it. He keeps going on that there's a little thing called an ocean between the US and Europe and the US and Russia. And so he thinks that definitely it's European countries and those closest to Ukraine and the Russian border who should be picking up their spending on that massively. And it's going to be very interesting to see how a lot of those countries and very much the UK included deal with it when he turns his attention to Ukraine. And what seems to be, we don't know when he'll do it, but it seems pretty clear what he wants to do is create a ceasefire and freeze things where they are at the moment and then start negotiations, which will almost certainly include Ukraine having to give up some territory. And if European allies of Ukraine and the UK don't agree with that, Trump's going to be very clear about who can pick up the bill to continuing to support Ukraine's defence. And I think that's going to be one of the bigger diplomatic, international diplomatic issues that the US and the UK are going to have to face off on. I think.

Speaker 4: What do you think, Chris? I think absolutely. It is going to be a really interesting conversation, that isn't it? Not least because, you know, when I speak to folk here and I think this will then play into some of the other issues that we're likely to hear from the White House. The UK political debate right now is obsessed about the generation of economic growth and creating the conditions where the economy has the greatest chance of growing. And we're going to hear from the Chancellor, Rachel Reeves, in a big speech on Wednesday. You know, and in that context, when you throw into the mix, for instance, the threat of tariffs, import taxes being thrown around by the new president, a big part of the argument you're going to hear, not just from the UK, but from others in his direction, is one where if other ideas coming from the White House could restrict economic growth, this is where the whole percentages thing breaks down a little bit. Then you end up with economies that are smaller, that are then allocating because of that percentage, a smaller proportion, sorry, a smaller real amount to defence, even if they are hitting a particular percentage indicator. And where does that leave the combined forces of NATO? So, yeah, I mean, I think it's going to be absolutely central to that conversation. But you know what? That is one of the areas where this conversation would have been happening anyway, I think. You know, imagine an alternative scenario where it wasn't President Trump. Okay, you might not have someone bandying around 5%, etc, etc. But that broader argument you'd be hearing from NATO members around the need to spend more than has been spent for a generation on defence, I still think would be very live.

Speaker 2: And Sarah, this weekend, you guys in the US, well, I mean, everyone everywhere, because we all witnessed it, had a real time example of Trump's use of tariffs, of import taxes as a tool to get his way. And it was getting his way with the South American nation of Colombia.

Speaker 3: Yeah, he certainly did. I'll run you through very briefly what happened. Because, as you know, the other thing that Donald Trump is very, very focused on alongside economic tariffs is immigration and trying to deport, well, ultimately millions, he says, of undocumented migrants who are living in the United States. And so there's been an effort in the last few days for customs officials to go and round up as many illegal immigrants as they can find and forcibly deport them and actually using military aircraft to do it. And the staging of this is all part of the optics of the theatricality of it. Donald Trump wants to be seen to be doing this as well as actually achieving it. So there was a military aircraft with migrants on it who had been handcuffed and treated fairly uncomfortably, was trying to land in Colombia to return Colombian natives. And the Colombian government were angry enough about this, about the manner in which they were being treated, that they refused the aircraft the right to land. And that created a bit of a diplomatic spat. Donald Trump was very angry. And he said that if this was the case, he was going to impose 25 percent tariffs or income taxes on all the goods coming from Colombia into the United States. That's a lot of coffee and it's a lot of fresh flowers as well, actually. It would really hurt the Colombian economy if he was to do that. And they were also talking about refusing visas and credentials for various Colombian officials, all sorts of diplomatic tools. And actually, Donald Trump ended up getting his way really quite quickly, because all he has to do is post this threat on social media. And before you know it, there are lots and lots of very urgent conversations going on. Whether the Colombians got any concessions or not, we don't know yet. But what we do know is that they have agreed that these military flights from the U.S. can now land. So without having to actually impose any of these tariffs, Donald Trump used his bully pulpit and his America first policy to get exactly what he wanted, it would appear, from the Colombians.

Speaker 2: My two observations from the weekend, which are probably less good quality than you two's observations because you're closer to these things than I am. Number one, on tariffs. So Rachel Reeves, chancellor, when she was on Laura's show on Sunday, was saying, oh, she didn't say these exact words, but she was sort of suggesting that the UK can relax a bit when it comes to the threat of tariffs. Because actually, in quite a few years, the U.S. has had a trade surplus with the UK. In other words, the U.S. has sold more to the UK than the UK has sold to the U.S. And quite often, Trump is talking about tariffs as a way of balancing trade deficits, where other countries sell more to the U.S. than the U.S. sells to them. So meaning that maybe the U.K. wouldn't be quite in the same position as some other countries. But what the Colombia example shows is that was to get an objective. That was as part of a negotiation. It wasn't to redress a perceived trade imbalance that Trump saw between the two countries. So maybe the U.K. isn't as potentially as exempt as Rachel Reeves might like us to think from that interview with Laura. And the second thing is, I watched over and over a clip of Trump having a town hall meeting in Los Angeles with the Democrat mayor of Los Angeles. And she's had quite a lot of criticism on social media for her handling of the fires. And it was also with a group of homeowners at this meeting who had lost their homes and were waiting to get access to their properties to demolish them and start rebuilding. And Trump was just there. He was like more like kind of like Robert Kilroy Silk from those old chat shows, those old audience shows where he was like sort of moderating between the angry people in the audience and the mayor of L.A. And he was just chatting away. And he was very chilled. And he was arguing with people and bringing people in and arguing with her. And it was just so relaxed. And he was just very, very loose. And I just think now that he, with his re-election, has been re-elected in a sort of with more legitimacy than they had the first time round, that will put a lot of pressure, I think, on politicians in other countries to be looser and to put themselves kind of in harm's way a bit more with members of the public and their opponents, because Trump just does it so naturally. And he's setting a new baseline for what political communication is.

Speaker 4: Totally. I was talking to a cabinet minister earlier on today who was saying exactly that. They'd watched on telly the other day when the president was signing all of those executive orders and making theatre out of it, political theatre, and throwing pens to supporters and all that kind of thing. And this person was saying that was theatrical delivery in action. You know, you voted for X and here's me starting to deliver X. Now, maybe you can do a bit more of that in a presidential system than you can in a parliamentary democracy. But my goodness, this person was saying that is a lesson you've got to take away. When we're so used to being able to order the delivery to our door next to no notice, in other words, that sort of age of instant gratification and the politicisation of that via his mechanism of communication is really, really powerful. It might not be to everyone's taste, but it's really, really powerful.

Speaker 3: Chris, that's absolutely fascinating that people in the UK are talking about it. Could you imagine four or eight years ago, any politicians from anywhere in the political spectrum saying we need to be emulating Donald Trump more? I mean, it's practically unthinkable. And he is a master of this, of creating these moments of political spectacle, of capturing everybody's attention. And also when it comes to what he was doing in the past week, very performatively doing things that his supporters want him to, that are maybe not the most significant things that his administration could possibly get around to, but demonstrate some level of delivery. And whilst maybe it's going to be harder to, for instance, tackle inflation, get prices down, some of the big things that he's promised. But if he's with a great flourish signing an order to release all of the January the 6th rioters, then that grabs everybody's attention whilst he's doing it. I think a lot of other politicians in other countries would struggle to do it with his level of panache and would just end up looking rather gauche and embarrassed as a result. But that they're even talking about wanting to try and emulate him, it's just that shows you how much he's rewriting the rules globally, doesn't it?

Speaker 2: Now, the stock market values of lots of big American tech companies plunged on Monday. And the reason they plunged was the news that there is a new artificial intelligence model on the block and it comes from China and it can pretty much do the same things that open AI and Microsoft and Google's and Meta's models can do, but much, much more cheaply because it uses much less powerful and many fewer computer chips. And this raises big questions about what road AI is going to go down. Will it go down a much cheaper, much more available route, which we hadn't expected it to, or will it stick on the road it was on, which was big American tech companies delivering it with very, very expensive tech? This is absolutely fascinating. It feels to me like we've reached a very 21st century fork in the road. So let's speculate about what's going on and what has led us to this fork in the road. I'm joined here in the studio by two of the finest human minds in the BBC. It's our AI correspondent, Mark Chislak. Hello, Mark. Hello. And Faisal Islam is here. Hello, Faisal. Hello. Faisal is a tech bro from Davos. Yes. I am.

Speaker 1: You've got your gilet underneath your jacket. Yeah, maybe I have to take the jacket off.

Speaker 2: I'm a bit confused. You're not up a mountain anymore.

Speaker 1: I should ask an AI to tell me how to dress.

Speaker 5: We'll come on to that later. It would say like this. It would say a gilet is absolutely the uniform that you need to realise that you're wearing.

Speaker 2: Well, the AI gets it wrong a lot. Anyway. And Mark, you've also come dressed as, like, you've come as Steve Jobs with your blue neck.

Speaker 5: Either that or a 60s cat burglar.

Speaker 2: Yeah. Anyway, right. Let's do the news. Our fashion podcast is short-lived, thankfully.

Speaker 5: Mark, what is DeepSeek? So DeepSeek, a Chinese AI company that's come up with a large language model, if you like. It's the sort of thing that we see from ChatGPT. It's caused a huge fuss over the weekend because it stormed to the top of Apple's free app downloads, the App Store charts, unseating ChatGPT. It's made the markets get a little bit jittery, which I'm sure you'll talk to us about. And the reason it's made the markets get a bit jittery is supposedly it's cost a really small amount of money in comparison to other AI models to train and to develop. It's apparently cost six million dollars to develop. Now, to give you some sort of sense around what that means, some of OpenAI's models have apparently cost in excess of a hundred million dollars to develop. So this has resulted in the value of quite a few tech companies taking quite a big hit on the markets today because there's a thinking, perhaps, that we're in a bit of an AI bubble, that these companies are overvalued, that the investment that we're seeing in them isn't really, you know, it's not representative of the value that we're going to see back from them in the short term.

Speaker 2: Yeah. So lots to talk about there. But just to expand on the market reaction, is this basically lots of traders going, oh, hang on, OpenAI, Microsoft, Google, you guys were about to spend billions and billions and billions. Maybe you don't need to spend. Is that what it comes down to?

Speaker 1: Yeah. At the very least, it raises very serious questions about an assumption about the way the world was going to work, which was these models would be essential for every corner of the economy, society, the government. Therefore, amazing amounts of compute power would be required, therefore, amazing amount of very advanced chips would be required. All of the companies in that supply chain saw their share prices go up and up and up and up and up.

Speaker 2: Particularly one called NVIDIA.

Speaker 1: Well, NVIDIA is used to specialising in gaming chips, and it turned out that that was best, you know, a huge application for that was in AI. And they, I mean, their share price at one point very recently in the last hour was down by 17%. It's lost nearly a fifth of its value. I think it's lost more than the value of Exxon, the oil company. So this is quite a serious re-evaluation, potentially. In any event, people thought that some of these tech companies were a little bit bubbly, that they couldn't really sustain these values, and some sort of pinpricking moment we were waiting for. I think there's maybe more than that, though, because actually, I think what it says is If you go back a week ago, when you think about that picture of the tech bros next to the president, and then them announcing half a trillion for Stargate, this big project for AI, and it just looked like American tech supremacy was set for decades.

Speaker 2: But it was actually for four days.

Speaker 1: Well, and then suddenly, you have this, and then you said it had just come out from Davos. It was actually the end of the World Economic Forum, and people started literally running around, have you heard about this deep-sea thing? And it raises questions, and it's not just a one-off. Various bits of investment have gone into China, and there's a fundamental question here, which is, we assume that the Western companies, and Silicon Valley in particular, is just the best for this sort of stuff. But in AI, what matters isn't just the quality of the innovation. By the way, China's got thousands and thousands of PhDs, but it's also the quality of the data. And it turns out that countries like China, for various reasons, have got a lot of data going into a lot more, with less qualms about privacy, perhaps, than the West. So could that be a route to better AI models? Perhaps not in this case, but it just raises all sorts of questions about our assumptions, not just about tech, not just about the economy, but also about power in the world.

Speaker 2: Right. So let's do a comparison then between this new kid on the block and the ones that we know a bit better, like ChatGPT. Have you tried them both out, both of you?

Speaker 5: Yes. Well, in certain circumstances, you could say that ChatGPT has got a little bit of a lead on DeepSeek. DeepSeek is apparently, and I've tried some of the maths applications and things, or tried to ask it some questions about maths and sciences and those kind of things, and apparently it scores better than OpenAI's apps will, or OpenAI's models will. Right. But when you start asking it questions about world events, when you start asking it questions about things that the Chinese government might be uncomfortable with, that's where DeepSeek starts to fall down.

Speaker 2: Oh, so if you ask it a question about the massacre in Tiananmen Square...

Speaker 5: It won't give you an answer, and it'll try and change the subject as well quite quickly. Right. It tries to change the subject, says it doesn't know about those sort of things, Tiananmen Square, the treatment of Uyghurs, Taiwan, a whole raft of questions that the Chinese government might be uncomfortable with, it will not respond to.

Speaker 2: And are they the same security questions around it that there are around TikTok?

Speaker 1: Well, if the Americans are going to ban TikTok, then you'd imagine they'd have more than an arched eyebrow about where this could go. I mean, I tried, I didn't try it myself. I outsourced this out to my kids, just to compare side by side. And of course, they're not using it to search geopolitical events of any description. It's like, rather, it was a rather cute question, what was it? How did whales get in the sea? And please, can you teach me algebra? And certainly on the second one, again, I can't, you know, for an 11 year old child reports to me that it's sort of a big slain...

Speaker 2: This is not industry standard benchmarking.

Speaker 1: But there was a way in which it explained things and reasoned and went through the sequence of trying to explain something complicated that they preferred to what they saw from Chai Chitpati. Now, these are all moving targets, these models, they're all getting better and better and better. But the key thing is not, is that I think, whatever you argue, it's kind of up there, right? But the key thing is, is it's so much cheaper in terms of the cost of using it and of deploying it.

Speaker 2: If you believe, if you believe what were their claims... That was gonna be my next question, because, okay, I'm not an expert in this. But very quickly today, I started seeing all these people raising questions about whether it was as cheap as they're claiming. Is it as cheap as they're claiming? Or why are there so many questions about it?

Speaker 5: Yeah, we don't know. Six million dollars sounds, a lot of industry experts have said that sounds like it's a made up number, or at the very least, it's a very massaged number. Because the numbers that we've seen so far, and even when we talk to some of the most innovative AI developers in the world, how quickly and cheaply they might be able to produce something, they've not said they'd ever be able to produce a model at a cost like that. So there are some big question marks over those sorts of numbers. But we can't forget that they've opened sources. So that means that they've just thrown it out there. Oh, yeah, just explain a bit more about that. So that means it's publicly available.

Speaker 2: I could just copy and paste all the code behind it if I wanted to.

Speaker 5: Yeah, you could use it. You can plug this into your own software if you want to, which is something that OpenAI hasn't done. Theirs is behind a wall. You can't do that. You can pay them to get access to it. And in some respects, you'll have to pay a small amount of money or a smaller amount of money to license this DeepSeq technology to do the same thing with your own software here. So the amount of money that they're asking in comparison to OpenAI is pennies.

Speaker 1: And I think this is why Mark Andresen, who's a really kind of noted Silicon Valley investor who's very close to the new administration, has called it a Sputnik moment. So that's obviously a reference to the Soviet Union launching their first satellite.

Speaker 5: I thought that was a... It's partially right, because the Soviet Union got the first satellite into space. They got the first human into space. First dog? First dog into space. In the space race, the Soviet Union did all of the big firsts, and it was really on NASA's shoulders to get a big win in there. NASA only really managed to get a man on the moon first. That was their big thing. So the Soviet Union was the sort of the preeminent power in the space race until the moon landed. The difference is now that the United States is currently the global AI suit.

Speaker 1: Yeah, they've got the dog in photographic space. My only point about this is, I mean, clearly, Mark knows more than me about the tech stuff here, but these people are taking it seriously, and they're taking it seriously as a moment of competition. They're taking it seriously that China has enough to compete and has enough to potentially collapse the cost of this. Now others are saying, well, if they collapse the cost of running, of making effective these chips, that shouldn't necessarily be bad for all of these companies whose shares, and it isn't just Nvidia, by the way, there's a whole sweep of shares. It's affecting the price of US government bonds right now. I wouldn't call it a stock market crash, but like this is, it's a serious move in the stock market at a time when the president is obsessed with the value of the US stock market. So let's see what he says about this, which he said nothing so far. But I think it's a big moment. It's what you call a black swan. We should have expected this. It was always US versus China. How they respond, and you mentioned TikTok, how they respond, given that they've banned TikTok to this, how we respond, what's the cybersecurity command going to say about whether this sort of stuff should be downloaded, about chucking data into that? I'm sure if it's school kids asking how to teach algebra, maybe that's okay, but it's amazing to think that it's the number one on the apps.

Speaker 2: I'm always interested in what the tipping point is, because it sounds like from listening to you two, it was there for people to find. Yeah, it's been there since the beginning of the year. Exactly. So what is it about the last three or four days that means it's now become a global talking point? What happened? People have noticed it.

Speaker 5: As simple as that. Yeah. And the idea of it being open source has made huge ripples. And I think as far as I was saying, the trajectory, everybody assumed the economic trajectory of pretty much most of the Western democracies was now deeply entwined with AI. Our government just a couple of weeks ago was talking about how it was betting big on AI. The Americans have been talking about their big bets on AI. So it looked like things were going in one direction, and this has properly thrown a cat amongst the pigeons by saying, oh, hey, okay, here's a totally different way of thinking about this. But it's also, there's an inspirational thing to take from this as well, that for countries that aren't already the dominant AI superpower and have all of this investment and have the ability to grab hold of this, you know, tens of billions of dollars worth of investment, it's inspirational for them, because it means that you can start thinking in more innovative ways. You can start thinking, you know, mother, you know, so what is it? What's that say? What's that phrase? The...

Speaker 1: You just give us a clue.

Speaker 5: I've forgotten it.

Speaker 1: Necessity is the mother of invention.

Speaker 5: That's it. Necessity has become the mother of invention, not being able to get hold of the investment or perhaps not being able to get hold of the technology has led to this happening.

Speaker 2: One of the reasons this firm in China had to go down the route of using less powerful chips was because the more powerful chips weren't available because of US controls on

Speaker 1: the market. This is a critical part of the story. Okay. So take a step. This is not even a tech firm in China. China's got brilliant tech firms. Some of them are embargoed by the West. This was a hedge fund. And he was doing it in his free time. And this was a side project. Having used AI to bet on the markets, they then, uh, they use this to develop a large language model. Now, a lot of this was built, Mark mentioned open source, was built on some of the open source developments in America and they're getting better and better and better. But it is extraordinary to think that this was a side project. One of the things I was shown at the World Economic Forum last week was when these things get totally intuitive and you can just chat to them like they were there on the step on the chair. And I was demonstrated something like this from Google Astra. And I was I was totally blown away because I was a bit sceptical about this. You do quite naturally start to trust it far more than you could ever imagine trusting a search.

Speaker 2: Right.

Speaker 1: You know, and then you wonder, and I don't know what the answer to this question is a

Speaker 5: lot of smoke and mirrors in a lot of those just talking about the psychological. Yeah.

Speaker 1: But we can envisage a situation where the quality of the output of something like chat GPT gets so good that it becomes conversational within the next three years. Absolutely. Yeah. Okay. So when that happens, your ability, the amount of information you will then give to that thing, like, you know, it's very difficult to restrict it. And imagine that, you know, you let your, you know, younger relatives, people having relationships with chat. Well, I mean, I wasn't already already doing it. We don't need to even go there. People are already doing that.

Speaker 2: Then, of course, there's the job implications of you. Well, you never need to hire another call center worker ever again.

Speaker 1: No, no. But I'm thinking about the business model item. I'm thinking like, like how in the past we started off searching and it was pure maths and algorithm. What's the most relevant? And then the advertisers sort of got in and it was slightly changed. What came the top of the top of the rank? Now, if that person, if that AI that you can actually converse with, which is like close, I would say, if that if that happens, and in fact, its answers are being biased by who's paying the most.

Speaker 5: But we're very good. We're very good as a species. It's spotting something that is inauthentic. Yeah, but they are very good. We are. We are very good. We are very good at starting to switch off when something gets inauthentic or when it becomes almost obvious that they're selling to us or that we're being directed towards

Speaker 2: I look forward to you two disagreeing about multiple times the next few years, but just the last one last note for you, Faisal. I mean, if you think back to the government's AI strategy two weeks ago, and then Rachel Reeves has got this big growth speech on Wednesday, like this is a government that moves at kind of like 20th century pace and actually the technology and the world and the markets are moving at 22nd century pace already. How do those two things coexist?

Speaker 1: I noticed a significant uptick in the pace at Davos last week, which was being done with the world trying to impress the world's investors. And some of the ideas they had was suddenly they were like they were coming from a different stream.

Speaker 2: Like ministers you mean?

Speaker 1: For ministers, yeah. And some of their advisors as well. So for example, they've quite quickly moved towards this position on trying to wave through people on visas in the AI industry and in pharmaceuticals. One way of reading that, you don't hear the government say this, but certainly from the pharmaceutical industry, for example, is, oh, look, something might be happening in America where pharmaceutical industry might not be quite the flavour of the month, shall we say.

Speaker 2: So there might be an exodus.

Speaker 1: Yeah. Well, there may or may not be. Or maybe there'll be a marginal one. Or maybe people think, maybe I've got to look somewhere else. Make sure you're ready to kind of sort of wave that human capital that's vital for growth in as quickly as you can. Are you starting to see slightly more nimble speedboat-like behaviour as regards trying to grab that growth? Were they ready for this in real time? Evidently not. Few people were. We've seen from the reaction in the stock market, they're going to have to get a response really quickly. And this is an extraordinary dilemma. I don't know what the answer will be. Do you say, like I suspect the Americans will get to, although Silicon Valley seems okay with it, do you stop this? Do you shut this down? Do you ban it, like TikTok? Or does Britain say, you know, this is actually transformative for the world. We need to sort of have a slightly middle ground. Don't know the answer to that question.

Speaker 2: We should say on the TikTok ban in America, it's got quite confused, hasn't it? So Congress passed a law that the courts kind of gave a thumbs up to, but then Trump came in and said, I'm going to give myself a couple of months to decide whether to implement this or not. Brackets, I'm not going to implement it. Because I love TikTok. No, you're absolutely right. Yeah. So it's a bit more confusing than that. Thank you very much. Thank you. Mark, thanks to you too. Thank you.