{"text":[[{"start":0,"text":"This is an audio transcript of The Economics Show podcast episode: ‘Is China catching up with the US in AI? With Kyle Chan’"}],[{"start":0.5,"text":"Soumaya Keynes"}],[{"start":13.1,"text":"There are two major players in the global race for AI supremacy: the US and China. Few dispute that the US is the technological leader, but increasingly the Chinese are catching up. On July 17th, the UK’s AI Security Institute said that the gap between Chinese AI models and leading US models is narrowing faster than previously thought."}],[{"start":36.76,"text":"In 2025, Chinese developers were six to 10 months behind US rivals. Today, as my FT colleagues have been reporting, the gap is as little as four months. It’s got to the point where the Chinese are considering turning the tables on the west. For years, they have been suffering western export restrictions on AI chips and technology. Now, the Chinese Ministry of Commerce is considering tightening export controls on its own AI tech."}],[{"start":63.26,"text":"So how is China’s AI strategy working? What do the Chinese want? And should the rest of the world be worried about it?"}],[{"start":63.76,"text":"[MUSIC PLAYING]"}],[{"start":71.04,"text":"This is The Economics Show with Soumaya Keynes. I’m joined today by Kyle Chan. Kyle is a fellow at the Brookings Institution, where his research focuses on technology and industrial policy. This is Kyle’s second appearance, which makes him an official friend of the show."}],[{"start":71.54,"text":"Kyle, welcome."}],[{"start":72.04,"text":"Kyle Chan"}],[{"start":87.52,"text":"Great to be here. Great to be back."}],[{"start":88.02,"text":"Soumaya Keynes"}],[{"start":89.64,"text":"Great to have you. OK, so on a scale of one to 10, how is China doing when it comes to developing frontier AI models? So, one is completely rubbish, 10 is spectacularly well."}],[{"start":90.14,"text":"Kyle Chan"}],[{"start":103.08,"text":"I think they’re at a nine. Not quite at the level of US frontier models, but there’s basically no one else in the game. It’s China versus the US, and the Chinese AI models are really hot on the heels of the American frontier models."}],[{"start":103.58,"text":"Soumaya Keynes"}],[{"start":115.52,"text":"And what would you say was the trajectory of that number? Where were they a few months ago and could they get to an 11?"}],[{"start":116.02,"text":"Kyle Chan"}],[{"start":123.24,"text":"(Kyle laughs) I think it’s been creeping up. I mean, looking back at the original DeepSeek moment a little more than a year ago, that was quite a leap in the numbers, say from like two or one to suddenly six or something like that. But now they are serious contenders, and they are getting global adoption and really starting to compete in the same spaces as the US frontier models."}],[{"start":123.74,"text":"Soumaya Keynes"}],[{"start":146.8,"text":"OK. Maybe an 11 one day."}],[{"start":147.3,"text":"Kyle Chan"}],[{"start":147.8,"text":"(Kyle laughs) Perhaps."}],[{"start":148.3,"text":"Soumaya Keynes"}],[{"start":151.8,"text":"I never said this was a serious scale. Well, look, more seriously, I want us to start off by talking about what China wants from AI. And we should distinguish here between the government and the private sector. So starting with the government, am I right in thinking that the government hasn’t actually been that vocal about winning the AGI or the artificial general intelligence race?"}],[{"start":152.3,"text":"Kyle Chan"}],[{"start":176.08,"text":"That’s right. The way I would put it is Beijing is super AI-pilled. They really see AI as this transformative, powerful technology, but they’re not AGI-pilled like folks in the US are. They’re not planning on some intelligence explosion that will utterly transform every single dimension of the world, much less destroy it all. So it’s a very different attitude and different approach."}],[{"start":194.08,"text":"And for AI, they see it as a way to turbocharge a whole bunch of other sectors of the economy. Some of the best examples are in the manufacturing and physical AI space. There we have a really strong effort to build out robotics, industrial robotics, humanoid robots, and to increasingly automate a lot of the manufacturing that China’s known for."}],[{"start":215.96,"text":"We now see, for example, humanoid robots at CATL battery plants helping to do battery testing or robots being deployed at electric vehicle factories. And it’s still early days, but I think you see a very concerted effort to try to deploy as soon as possible and scale up even when one might argue the technology isn’t quite ready yet."}],[{"start":236.06,"text":"And they’re trying to boost adoption. That’s really their number one goal. It’s about adoption and diffusion throughout their economy, and then ideally globally."}],[{"start":236.56,"text":"Soumaya Keynes"}],[{"start":540.13,"text":"Is there a coherent reason for why they’re not AGI-pilled?"}],[{"start":540.63,"text":"Kyle Chan"}],[{"start":541.13,"text":"It’s a good question. There are different reasons. One is that if you look at China’s efforts to catch up or become a leader in emerging technologies in other areas, they have a very pragmatic approach to a lot of these things."}],[{"start":541.63,"text":"They’re not sort of extrapolating to some blue-sky potential future that could be utterly different. They’re thinking about what is the next step. How can you make this one thing a little bit better? And hoping that will add up incrementally over time to something beneficial in the long run."}],[{"start":542.13,"text":"So yeah, there does seem to be a kind of pragmatism, also a kind of focus on the physical world. For Chinese leaders, it’s about making stuff, about applying technology to the real world. And so the idea of a digital god is just not something that resonates so deeply with leaders in Beijing."}],[{"start":542.63,"text":"Soumaya Keynes"}],[{"start":540.13,"text":"Is there a sense that the US has that covered? They’ll burn billions and billions of dollars racing for this AGI thing, and we’ll just clean up afterwards and make sure that we get rich actually implementing it?"}],[{"start":540.63,"text":"Kyle Chan"}],[{"start":541.13,"text":"Yeah. Actually, I do think that folks in the Beijing policy circles are kind of scratching their heads watching what’s happening in the US. I think they see the enormous investment being poured into data centres, something on the order of like $1tn if you look at the projections for next year, and they’re kind of wondering what we’re up to, and maybe we’re wondering what we’re up to as well. But yeah, I think they have a very different attitude towards this technology, and I think they’re hoping to win a totally different race. A race that’s actually about deploying and making the greatest use of AI. That’s why if you look at all their policies, whether it’s their AI Plus action plan, whether it’s their Global AI Governance programme, they’re really focused on trying to make AI useful in the here and now."}],[{"start":541.63,"text":"Soumaya Keynes"}],[{"start":542.13,"text":"Has that been changing at all? Has the government become more enthusiastic recently, or has it been a constant level?"}],[{"start":542.63,"text":"Kyle Chan"}],[{"start":543.13,"text":"So China has been taking AI actually more seriously over time, and I think that’s best captured by Xi Jinping giving the keynote speech this year at the World AI Conference in Shanghai, where he’s talking about Chinese AI leadership, trying to portray China as not just a leader in the technology, but in sharing this technology with the rest of the world."}],[{"start":543.63,"text":"And so I think it’s become increasingly central to what they’re trying to do. It was maybe one of a number of different areas of focus, and now it’s become almost the meta technology that underpins all their other technological ambitions."}],[{"start":544.13,"text":"Soumaya Keynes"}],[{"start":544.63,"text":"OK, so President Xi Jinping is into it. How much is there a concern that this is going to lead to widespread job loss, right? That could generate social discontent, something that the government has been quite concerned about historically. Are we seeing any signs of some kind of safety plan?"}],[{"start":545.13,"text":"Kyle Chan"}],[{"start":545.63,"text":"It’s interesting, China has sort of swung back and forth like a pendulum when it comes to accelerate or regulate AI. Actually in the earlier days, they were much more on the cautious side because of political risks. They wanted to really tightly control the political content, make sure these AI models don’t say things that aren’t supposed to be said about Taiwan or other sensitive topics."}],[{"start":546.13,"text":"But then, I think partly because they saw the rapid progress we made in the US, they kind of shifted more towards “Actually, let’s hit the accelerator pedal here.” And they’ve really been trying to go all out in boosting the entire AI stack: the models, the infrastructure, the energy, we can get into this."}],[{"start":546.63,"text":"But now, it’s yet another perhaps turning point where China seems to be doing very well on the model performance. They seem to be doing very well on the diffusion, but now there’s all sorts of questions about job loss and about even risks inherent in the technology itself. Terms like loss of control, I think, are becoming more common in the policymaking discourse."}],[{"start":547.13,"text":"So I think the attitude is shifting, and I do think part of that is because they’re watching very closely what we in the United States are doing."}],[{"start":547.63,"text":"Soumaya Keynes"}],[{"start":548.13,"text":"OK. So I’ve asked you what the central government wants when it comes to AI, which is widespread adoption, greater productivity, greater growth. What about local government?"}],[{"start":548.63,"text":"Kyle Chan"}],[{"start":549.13,"text":"Yeah, so local governments are really important, and they’ve been jumping in to be, you know, one of China’s key AI hubs. We’re talking about Beijing, which I think most people might associate with being a stodgy capital, but they’re actually really important because they’re the home of some of the top universities like Tsinghua, as well as top Chinese AI labs."}],[{"start":549.63,"text":"Shanghai is trying to play this role as sort of like a global AI hub, and they were really the ones as a city that were putting forth this, quote unquote, world AI conference, right? It kind of gives you a sense of their lofty ambitions."}],[{"start":1008.46,"text":"And then you have Hangzhou, which is the home of DeepSeek, as well as Alibaba, and they have actually quite a number of really strong start-ups and tech companies and have really made a mark in the AI world. And then you have Shenzhen, kind of like the hardware capital."}],[{"start":1008.96,"text":"So each of these different local governments is trying to boost their own part of AI, and they are each sort of vying with each other to be the new capital of AI for China, if not for the rest of the world."}],[{"start":1009.46,"text":"Soumaya Keynes"}],[{"start":1009.96,"text":"So are there big state-owned enterprises that are major players in this space? Or is it really the private sector that’s pushing forward this AI surge?"}],[{"start":1010.46,"text":"Kyle Chan"}],[{"start":1010.96,"text":"It’s really a private sector-led thing right now. You have kind of the Chinese Big Tech companies like Alibaba, Tencent, ByteDance and they have, in many cases, not just AI labs, but sprawling tech ecosystems across ecommerce, social media of course for ByteDance, which is the owner of TikTok. And then you have the AI start-ups like Z.ai or Moonshot, which just released Kimi K3, as well as of course DeepSeek."}],[{"start":1011.46,"text":"And these are all private players, and they’re all trying to make money in this process. That’s really their goal. They’re trying to develop better and better models. They’re trying to integrate them, if they have an ecosystem, into their broader set of services, and they’re trying to carve out space, especially within the Chinese market. A number of them are also looking internationally, but a lot of the competition is there within China trying to win over users, both consumers and businesses."}],[{"start":1011.96,"text":"Soumaya Keynes"}],[{"start":1012.46,"text":"And are their business models basically the same as their US counterparts? So kind of a user fee or even kind of a cost per token?"}],[{"start":1012.96,"text":"Kyle Chan"}],[{"start":1013.46,"text":"It’s converging to a certain extent, but one big difference is that a lot of the Chinese AI companies have emphasised open source. That is, technically, they release their model weights so that anybody can download them, or you can go to other cloud platforms and run these Chinese models. So in that case, they’re not making so much money directly from the model itself in the closed API format that we’re used to in the US, where you’re paying per token or per usage."}],[{"start":1013.96,"text":"Now this is shifting, because the Chinese companies, they’ve also been investing a lot of money in their infrastructure, in their data centres, and in training these growing models. So they also need to recoup some of that money. A number of them have also gone IPO recently, and more are expected to come."}],[{"start":1014.46,"text":"So there’s pressure to commercialise, and I think if anything, we’re seeing a hybrid approach where they’re trying to still maintain some of their open-source strategy but look for ways to generate more recurring revenue, subscription revenue or just direct API fees."}],[{"start":1014.96,"text":"Soumaya Keynes"}],[{"start":1015.46,"text":"OK, that’s interesting because I’ve been reading reports in the FT about the Chinese authorities considering maybe restricting access of these models to international users. But what you’re saying is actually there could be more than one impetus to make these models less transparent, less open, the main one being commercial."}],[{"start":1015.96,"text":"Kyle Chan"}],[{"start":1016.46,"text":"Yeah, so what’s happening here is there are a couple of things going on at the same time. On the commercial side, there are just plain economic reasons for wanting to shift to a more closed approach like the US."}],[{"start":1016.96,"text":"You can make a lot more money, maybe not become a trillion-dollar start-up, but at least a little bit more. Then there are some national security potential risks from the policymakers’ standpoint in Beijing, and this is where I think we’re seeing some of this discussion around potential restrictions or export controls on Chinese open-source models going out to the rest of the world."}],[{"start":1017.46,"text":"So, you know, I’ll caution and say that it’s still very early, and there’s a lot that we don’t know. But there have been some interesting discussions around potential data leakage. Some worries that in the model weights themselves, if you download, say, Kimi’s latest model or Z.ai’s latest model, that you might be able to figure out something in the data that is sensitive within China because that’s part of the training data that goes into these models."}],[{"start":1124.2,"text":"So I think going forward, what will be really interesting to see is whether there will be more restrictions and what kind of restrictions Beijing might consider imposing, whether that’s stopping or limiting open-source models to some extent, whether that’s trying to restrict who can access these open-source models. That’s a much more difficult technical problem to deal with, or whether there’s gonna be a broader shift towards closed models."}],[{"start":1124.7,"text":"Soumaya Keynes"}],[{"start":1125.2,"text":"I wanna ask about how this is framed in China, because in the west or in the US, there’s really this framing that it’s a US-China race, right? It’s very adversarial. Is that the same in China, or is this a case of “You’re so vain, you probably think this song is about you”?"}],[{"start":1125.7,"text":"Kyle Chan"}],[{"start":1126.2,"text":"It’s a great question because on the one hand, the Chinese AI labs and Chinese AI researchers watch what’s happening in the US very intensely. They’re always keeping track of the latest developments, the latest models. But at the same time, though, I think when it comes to day-to-day competition and strategy, they’re really focused on trying to create their own business and get their own market share, and that’s probably their number one goal. And so they’re trying to cater especially to the Chinese market as much as possible."}],[{"start":1126.7,"text":"Soumaya Keynes"}],[{"start":1127.2,"text":"Are these Chinese companies trying to serve US customers? Or are they really focused on the Chinese market or the rest of the world?"}],[{"start":1127.7,"text":"Kyle Chan"}],[{"start":1128.2,"text":"The US is sort of the final frontier for Chinese companies, not just in AI, but for a whole bunch of tech sectors because they’re competing with really, really powerful American competitors who are battling it out on their home turf. And they’re dealing with not only the sort of brand risk, the “Made in China” association with cheaper, lower quality, but also the national security and compliance risks that are associated with Chinese technology. They’re trying though, I mean, you can see how active they are on platforms like Twitter or in actually doing interviews with the media."}],[{"start":1128.7,"text":"It’s surprising to me how outward-facing they are, and they’re trying to get attention because, at the end of the day, their open-source strategy, a lot of their efforts are about trying to get people to take their model seriously, try them out, and maybe start to integrate them into their workflows."}],[{"start":1129.2,"text":"Soumaya Keynes"}],[{"start":1129.7,"text":"So when it comes to investment, right, the US is in a league of its own. But what are the areas in which actually it is acting as intensely?"}],[{"start":1130.2,"text":"Kyle Chan"}],[{"start":1130.7,"text":"Yeah, so I think perhaps the single dimension where the US and China are competing the most intensely is on the model performance themselves, on trying to get better and better on the key benchmarks, on trying to build out some of these key features, like being able to do agentic tasks or build out coding capabilities."}],[{"start":1131.2,"text":"You definitely see a lot of convergence there between the latest GLM model from Zhipu or some of the multimodal models coming from Moonshot or ByteDance. Those are trying to emulate or compete with some of their US counterparts more or less directly."}],[{"start":1131.7,"text":"Soumaya Keynes"}],[{"start":1132.2,"text":"OK, but are there any areas within the kind of AI stack or supply chain where actually the Chinese are more impressive than the US?"}],[{"start":1132.7,"text":"Kyle Chan"}],[{"start":1133.2,"text":"Energy, for sure. China’s able to build out energy at a very rapid pace. Just last year alone, they built out something like the equivalent of 40 per cent of the entire US power grid in a single year, and a lot of that’s renewable. Just last year, 80 per cent of the new energy capacity in China was renewable energy, especially solar and wind."}],[{"start":1133.7,"text":"So they’re already able to scale up energy for non-AI-related reasons. So they’ve invested very heavily in energy, and that has actually come to help now in this new era of AI, where that is a potentially major limiting factor in the US, being able to not just build the data centres but power them. And now in China, they don’t face such a strong bottleneck in that area."}],[{"start":1134.2,"text":"There’s still some nuance there. I mean, you can see that some of the data centres being built in the major coastal areas where the urban users are, those ones are starting to strain the local grid. But overall, you just have a much greater capacity for building out new power in China."}],[{"start":1134.7,"text":"Soumaya Keynes"}],[{"start":1135.2,"text":"What are the other constraints stopping China from reaching its full AI ambitions?"}],[{"start":1135.7,"text":"Kyle Chan"}],[{"start":1136.2,"text":"There’s a whole bunch. So in addition to the chips, which are export controlled, China can’t access the best Nvidia chips, although they have some ways to access some of them, and the fact that Chinese chips are getting better and better, especially for inference, but still lag substantially. That’s one major bottleneck."}],[{"start":1136.7,"text":"Another major bottleneck is data. The US has a very robust data economy, as it were. You have companies that specialise in hiring a bunch of human experts and having them annotate data sets, label them so that they can go into training models. And that data is really, really precious because everyone’s already sort of vacuumed up all the internet data that’s out there, and there’s a lot of variation in quality."}],[{"start":1137.2,"text":"But high-quality data is something that China’s trying to build up more, and you can see that. There are government-led efforts to try to build high-quality data sets. There are industry-specific efforts, especially in healthcare and biotech, where you have, say, doctors labelling, say, X-rays or MRIs and trying to feed that into Chinese AI models. But data is another major bottleneck."}],[{"start":1137.7,"text":"And then you have sort of like the broader AI infrastructure supply chain, which is much more than just chips and the data centres. You have the power equipment, like I mentioned earlier, transformers. You have also network equipment. You have cooling systems. Some of that China is very good at, but in other areas, they’re still dependent on other countries and other companies."}],[{"start":1138.2,"text":"Another area that’s kind of related to that is memory. So in addition to the GPUs, you have the memory chips as well. So I could go on. The AI stack is quite vast, and not every country has a lock on, you know, the key layers in that broader AI stack, even China, which would claim to have, you know, such a broad base of industrial capabilities."}],[{"start":1138.7,"text":"Soumaya Keynes"}],[{"start":1139.2,"text":"OK. Well, let’s go to a break now, and when we get back, I want to talk about what policymakers elsewhere in the world should be doing about China’s AI prowess."}],[{"start":1139.7,"text":"[MUSIC PLAYING]"}],[{"start":1140.2,"text":"We are back from the break. Kyle, how do you think that US policymakers are feeling right this minute about all of this Chinese AI progress? Are they relaxed, having a nice time at the beach?"}],[{"start":1140.7,"text":"Kyle Chan"}],[{"start":1141.2,"text":"Yeah, so Washington is worried about several things at once. One is these more powerful Chinese models eating into the market share and revenues of American AI companies."}],[{"start":1141.7,"text":"So keep in mind, American AI companies have been investing a ton of money, and they need to recoup that investment through subscriptions and API access. And to the extent that companies, including in the US, might substitute for some of that OpenAI or Anthropic usage with, say, a Kimi or DeepSeek, that will eat into some of that potential revenue that could have been earned."}],[{"start":1142.2,"text":"Another major concern is about a broader sort of technology leadership on AI and the idea that the US should be the leader in this space and should have something to export to the rest of the world. There’s been this talk in the Trump administration about exporting the American AI stack, and that includes the AI services as well as the chips and the hardware that goes with it."}],[{"start":1142.7,"text":"And then another concern is on the cyber side about whether not only China will have these AI-enhanced cyber capabilities, through open source, will allow essentially everyone to have access to a Mythos-level cyber capability. I think that has a lot of folks in Washington concerned as well."}],[{"start":1143.2,"text":"Soumaya Keynes"}],[{"start":1143.7,"text":"Do you think that the US should be taking a leaf out of the Chinese playbook and focusing a bit more on adoption within the US economy?"}],[{"start":1144.2,"text":"Kyle Chan"}],[{"start":1144.7,"text":"Absolutely. I think right now we have basically a highly leveraged bet on AGI, and we also have growing backlash to AI from regular people. You see the backlash to data centres but also to the technology itself. You have young people struggling to find jobs and saying that AI could be a potential reason for that."}],[{"start":1145.2,"text":"Right now, what would really make the biggest difference in the US is trying to integrate AI more directly into people’s lives in a way that’s actually productive and beneficial to them, where they can feel the benefits more directly."}],[{"start":1145.7,"text":"I think right now it seems like this is more abstract technology to a lot of people that might not be something that benefits them directly but could be a risk to their jobs or to their education or to other things."}],[{"start":1146.2,"text":"Soumaya Keynes"}],[{"start":1146.7,"text":"So there’s a lot of hand-wringing in the US. Is there also demand to work with the Chinese? I mean, surely there are lots of companies who just wanna use the cheap Chinese models for the basic stuff and then just pay for the US models for the fancy stuff."}],[{"start":1147.2,"text":"Kyle Chan"}],[{"start":1124.2,"text":"Yeah. There is actually a lot of interest and demand in the US for these Chinese models. Not only are they cheaper, but in many cases you can control them much more. You don’t have to be sending data to someone else. You can customise them for your own needs, maybe fine-tune them on your own data. So yeah, there’s increasing interest, and I think what’s interesting is to see the range of users in the US that are building on Chinese AI models."}],[{"start":1132.32,"text":"I know personally, you know, researchers, just people on an individual basis who are playing around with AI agents. Many of them are switching back and forth between US models and Chinese models because they don’t feel like they need the most powerful models all the time for their coding projects on the weekends."}],[{"start":1149.12,"text":"And then there are companies, especially outside of the larger US companies, that are cost-conscious, that maybe want to build a customer service chatbot, but also don’t need to have the most powerful model in the world to be running that. And then more recently, I think what’s been most surprising to me is to see even some of the largest companies in the US start to be worried about AI costs, given that they were leaning so heavily into trying to just boost AI usage sort of no matter what."}],[{"start":1175.28,"text":"That’s this whole phenomenon of token maxxing, where it’s like a race to see who can use more AI and somehow then you win. So what’s interesting is even for some of these companies that seem like they have limitless buckets of cash to be spending on AI, they also don’t want to be handing over a billion dollars a year to OpenAI or Google."}],[{"start":1194.16,"text":"So across the board you see growing interest, and I think probably most of all it’s the cost followed by the control that makes these Chinese AI models so appealing."}],[{"start":1194.66,"text":"Soumaya Keynes"}],[{"start":1204.16,"text":"OK, but there’s no question of, say, OpenAI or Anthropic going into partnership with one of the frontier Chinese companies?"}],[{"start":1204.66,"text":"Kyle Chan"}],[{"start":1213.84,"text":"I would be pretty shocked if that were to happen, given how fiercely they sort of compete with each other, or at least how much OpenAI and Anthropic see them as competitors."}],[{"start":1223.16,"text":"Anthropic in particular has quite a strong framing of Chinese AI. They recently had that blog post framing it as not just different AI companies trying to compete commercially, but a battle between democratic versus authoritarian AI, and what are the stakes if basically China wins on AI. So they’ve been leaning in the hardest in terms of restricting access in China to their own models, for example."}],[{"start":1223.66,"text":"Soumaya Keynes"}],[{"start":1248.31,"text":"OK, last question. There is going to be a meeting between US and Chinese policymakers in September where AI is going to be discussed. Is that just a talking shop? Do you think they’ll achieve anything?"}],[{"start":1248.81,"text":"Kyle Chan"}],[{"start":1258.88,"text":"So I think what’s ironic here is that both countries feel very confident in their AI progress. Trump likes to talk about America being the best, the leader in AI, and I think China is increasingly seeing itself as a major leader. And so that makes it easier to at least sit down at the table together and talk about AI, I think having that level of comfort. But I think that there are a number of thorny issues that will be really difficult to hash out."}],[{"start":1282.92,"text":"One is about the open-source question. I think one could imagine the US side of the negotiating table pushing China to do more to tighten their controls over open-source models. And for China, this is really central to their AI strategy."}],[{"start":1295.96,"text":"Another major area of dispute could be on the distillation issue, and the US might want to pressure Beijing to then in turn pressure their own AI labs to cut it out on distilling on US models."}],[{"start":1307,"text":"But then there are areas where they could actually make real progress and have pretty strong common interests, and one of those is trying to deal with non-state actors. I don’t think any country would want to have their own AI models enhance cybercriminal organisations or help with terrorist groups."}],[{"start":1323.28,"text":"So I think that this is probably an area where at least they could affirm that they have this shared interest and then start to see what would be the concrete steps towards building up, I don’t know, guardrails for AI models with regard to these sorts of malicious non-state actors. So there’s a lot that could be on the table."}],[{"start":1339.82,"text":"There’s a lot that they should be talking about. I do think that one really important thing is that everyone keeps their expectations pretty low, because I don’t think we’re gonna get some nuclear arms treaty for AI out of this next upcoming meeting. If they can start to talk, that’s actually already quite a big step up."}],[{"start":1340.32,"text":"Soumaya Keynes"}],[{"start":1356.76,"text":"OK, so expectations for this meeting, pretty low, but we wanna end the episode on a high note. So expectations for AI pretty high in China?"}],[{"start":1357.26,"text":"Kyle Chan"}],[{"start":1367.56,"text":"Oh, yeah. I mean, if you look at their stock market, there’s definitely been an AI sugar high, so we’ll see what happens."}],[{"start":1368.06,"text":"Soumaya Keynes"}],[{"start":1373.8,"text":"OK. Kyle, thank you so much for joining me."}],[{"start":1374.3,"text":"Kyle Chan"}],[{"start":1374.8,"text":"My pleasure."}],[{"start":1375.3,"text":"[MUSIC PLAYING]"}],[{"start":1375.8,"text":"Soumaya Keynes"}],[{"start":1377.84,"text":"That is all for this week. You have been listening to The Economics Show with Soumaya Keynes. If you enjoy the show, then I would be eternally grateful if you could rate and review us wherever you listen. This episode was produced by Mischa Frankl-Duval with original music and sound design from Breen Turner. Flo Phillips is the FT’s head of audio. I’m Soumaya Keynes. Thanks for listening"}],[{"start":1405.2,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1784984614_7026.mp3"}