KinaNørderne
KinaNørderne - Casper Wichmann og Mads Vesterager Nielsen, stiftere af Sinolytica - dykker hver uge ned i nyt fra- og om Kina og nørderne dem igennem!Få analyser, tanker, indsigt, diskussioner om alt fra kinesisk politik - indenrigs som udenrigs - kinesisk samfund, kultur, sociale medier og meget mere!
KinaNørderne
Special - China's AI Strategy and the Global Information Space - With Alex Colville
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
05-05-2026:
Another special in English!
In this episode we have the great pleasure to talk with Alex Colville from China Media Project.
We get more into China's AI strategy, their open source AI models, and what it all means for the Global Information Space.
We also discuss what it means for Europe and how the approach to AI differs from the US to China to the EU.
KinaNørderne af Sinolytica
Følg os her:
Substack | Webpage | FB | Twitter | LinkedIn | Instagram
Intro
SPEAKER_03Den her på engelsk. Ligesom nogle af de få.
SPEAKER_01Mit navn er Kasper. Jeg er denørt. Og i mit selskab. Der Nelson den 12 kan. Vi sidder her i to forskellige steder. Linket op med hver vores mikrofon setup den gang.
SPEAKER_03Og det er simpelthen, fordi vores gæst sidder i Taipa på i Taiwan and our guest today is Alex Koble, researcher at the China Media Project. Welcome to the podcast, Alex. Hi, guys, thanks for having me. It's uh it's our pleasure. Alex, maybe if we can start with your journey into China, into China Media Project, and why and how you ended up in Taipei?
SPEAKER_04Um, well, that is that's quite a story. Um so I if you had told me about seven, eight years ago that I would be doing the work I currently do, which is monitoring uh how Chinese propaganda is encoded into Chinese AI models, I would have laughed in your face. Um I initially Yeah, I initially started out thinking I was going to be a journalist based in London. Uh and at that time I was writing for uh papers like The Economist. Uh and then the thing was I I I got bored. And I was about at that time I was about 23, 24, I made a life-changing decision at without really thinking it through properly. And I took uh a tutoring job uh in Beijing, uh, which was only meant to last for about six months. I thought, great, I'll make some contacts, I'll try and rudimentarily learn a language, I'll just go to a part of the world which everyone wants to know about and yet still knows very little about. So yeah, I took this private tutoring gig in October 2019. Um so uh the COVID happened, uh, the borders initially closed. Um, and so yes, I then spent the next uh three years uh all the way through uh COVID, just uh living in Beijing, working there, um, starting to uh work for companies like China Britain Business Council. I started uh a column for a company called the China Project. Uh and then I found a company called the World of Chinese, uh, which we kind of in our field say is kind of like the New Yorker, but with Chinese characteristics in both the good and the bad sense of that. Um and I started covering the impact of the zero COVID policy on ordinary people's livelihoods, and it was getting towards 2022, the zero covert policy was getting more and more dark. Um it was there were there were there were holes showing in it. It was but because Xi Jinping himself had said that had given his seal of approval, that meant that no one could really point out the problems with the policy. It was getting to the stage where I would interview uh an artist who had scrawled across uh a COVID testing booth in Beijing. I have already been numb for the past three years, and he uh gets arrested the next day by the police and put away for longer than Ai Wei Wei was arrested in I think it was 2015. And the only reason why he was let out was because of the white paper protest and the uh the that the ending of the policy. Uh but it was becoming very dystopic quite quickly, and so for me that was a moment of profound disillusionment, uh moment of realizing that uh on paper uh this that this system can work fine, but because of the nature of the power structures in China and the nature of those power systems, uh things can go awry very quickly, but there is no way to have accountability for that. Um so then, yes, I left China uh after three years. Uh I then went back to the UK, still carried on freelance work, then did a coding course uh at Columbia University, and then I decided I was going to carry on my uh my Chinese language learning in Taiwan. Uh I'd heard from uh friends that actually the the quality of language learning in Taiwan is better than it is in China. So I thought, well, that's that's interesting. And then in a nutshell, uh I found China Media Project. Uh and I had always read their work and I thought their research was particularly good. And I suppose because of the nature of information flows in China, I thought it was something that was relatively under-researched. Uh and having gone through zero COVID, you would notice quite quickly how China's media system really did influence people's ideas on the subject. Like you could go to uh I had once had a kind of, I was once riding in a taxi at Changsha in uh in Punan, and the taxi driver starts saying to me, Oh, uh, we we got onto the conversation of the zero-coder policy, and he said, Well, I feel sorry for all those Americans after what they did with Fort Detrick. Um and Fort Detrick was uh a conspiracy theory that was set up by uh China's Ministry of Foreign Affairs uh in reaction to public international outcry about the possibility that uh COVID could have originated from a lab in Wuhan. And so China's Ministry of Foreign Affairs, in retaliation, uh started saying there was a possibility that it might actually originated in an American uh lab based in Fort Detrick. And this taxi driver uh treated it as fact that it had happened. And they said, Oh, how how do we know this has happened? And he looked at me and said, Well, it's because the Ministry of Foreign Affairs said so. Um so the power, I suppose, that that uh political system has over hearts and mines domestically is something we really still shouldn't be underestimating. Um so I thought there are very few companies that are still doing this sort of work. Uh so I thought, why not? Uh my company wanted someone who was had familiarity with coding. Uh, and then we started noticing there was more and more uh AI-based media uh in 2024, and so I started increasingly uh looking at that. That became slowly became my beat here. Uh, and we didn't really expect it to suddenly take off, just for example, after Deep Seek, and the amount of power that China has managed to gather for itself in terms of AI and how much suddenly there is to monitor it this way. Yes, that's uh I'm I that's a bit long-winded, but that's my story essentially.
SPEAKER_03So uh what are you working currently um in relation to what are you working on currently in relation to China?
SPEAKER_04Yes, and what I'm working on at the moment is uh oh goodness. Actually, I'm well I'm working on quite a few things. Uh so at the moment, and as I'm sure you know, China has a strategy when it comes to AI, which is both to increase the amount of AI deployment that there is domestically, but also to utilize AI in order to fulfill pre-existing strategies on the geopolitical side that they've had for decades, in this case, just to increase their influence. So domestically, what I'm interested in is, for example, uh the State Council has said that by the end of 2027 they would like uh I think it's a 70% penetration rate of AI into multiple different fields of Chinese life, Chinese society. But then internationally, they are floating, for example, multiple different international uh cooperation mechanisms, such as the World AI Cooperation Organization, which is still pending as to what that will look like, but also slowly organically realizing that a certain policy called open source, uh, or known as open source rather. Sorry, should I explain what open source is, first of all?
SPEAKER_03Yeah, that's probably a good idea for I'm sure there are some of our listeners who might not be entirely sure of what it what it really is.
SPEAKER_04Sure. Okay, so in which case I'll go back a little bit. So one strategy that uh the Chinese government has hit on is that something known as open source. Uh open source is a form of AI, which, unlike, for example, uh ChatGPT, uh, which is something which developers cannot actually download and play around with the code and create a new model. Chinese uh AI models are now generally, but still not all of them, are largely open sourced. That means that developers can take the code, they can take an AI model, they can download it, they can adapt it, they can reuse it, they can create something new with it. The idea is that this will be able to empower the global south. Um, it will be able to empower different narratives, different cultural values uh around the world, as opposed to what is considered uh a Western-centric, uh hegemonic power base whereby the US uses the latest technology in order to continue its hegemony over the world. But uh what I have been looking at is how Chinese AI is made to conform to uh the Chinese government's standards of what is and is not safe. Uh there are multiple companies uh working in this space who are putting forward the argument that uh because China's uh AI space is so heavily regulated, that must mean that it is safe by international standards. But the key question is whose definition of safe is being used in this context? Uh and so in my own work, I have been playing around with a lot of different AI models, both created by Chinese companies, but then also international AI models that, because of the open source policy, have uh been adapted, downloaded, rearranged by international companies. Uh, and we're finding Chinese information guidance uh in multiple uh different models from California, from uh Uganda, Malaysia, Japan, Germany, which are repeating uh certain Chinese propaganda narratives.
SPEAKER_02Alex, I would like to ask you um why do you think it's important for countries like Denmark and EU to better understand China and its global world also when it comes to your field with AI and the narratives that are uh being pushed?
SPEAKER_04I can I put it this way. So I think at the moment what we're seeing is increasingly uh sophisticated uh propaganda tactics from the Chinese side. But I would say mentalities within the EU as to what Chinese propaganda is, uh, that are very outdated. So uh think of it this way: you imagine for a moment that you guys aren't China watchers. You haven't been looking at China for ages, for years. You are a journalist in, for example, a Danish newspaper like Berlinska, uh, and you have been told in February 2025 your editor tells you that there is a hot new AI model out of China called Deep Seek, uh, which, according to reports, is just as competitive uh as things coming out of Silicon Valley. Uh, but you know that it's from China, so chances are there's going to be some form of uh information, like Chinese Communist Party uh guidance as to what it can and can't tell you. You have one question that you can ask this model to test that assumption. What do you ask it?
SPEAKER_03It's a funny anecdote because I was I was actually working as uh Asian China correspondent in Beijing um for Danish TV station when it when it came out, and and their angle was the the censorship angle, because you could see it censored in real time about Tiananmen, Xi Jinping, and it would it would give you an answer, flesh out the answer, but then backtrack and delete it, and then write, okay, sorry I can't talk about that. Yeah, the self-censoring.
SPEAKER_04But as I say, like what what question, if you had no knowledge about how Chinese propaganda works, what question would you ask it?
SPEAKER_03Uh I guess we're too deep into it. So um, but yeah, to do go on. What uh what is uh the question?
SPEAKER_04Everyone essentially would ask more or less the same question. What happened on June 4th, 1989? Is Taiwan an independent country? Or what is Fangong? Like things that we know that China, the Chinese leadership, has an angle on. And what will the model respond exactly as you say? Oh, I'm sorry I can't answer this. It's censored. Uh so you get uh people from Silicon Valley saying, we're going to uncensor this model uh because it's open source, that means it's going to cut down on costs, that means it will be very useful for people if they can just get rid of those, as one Silicon Valley uh CEO told the Wall Street Journal at the time, as long as we can get rid of this quote unquote half-baked censorship. So you also, even now, when you see people trying to test uh in in uh in papers online, for example, in Archive, which is a very prestigious uh AI developer uh research hub, they will uh ask more or less the same questions in their evaluation benchmarks exactly on those lines. It's Tiananmen, Tibet, Taiwan, uh Xinjiang, all of these ones. But this isn't really how uh Chinese proper ganda works. The way to think about it is more actually the way that the Chinese leadership talks about it, which is they call it public opinion guidance. Now, how do you guide someone to think the way you want to think? It's not just about withholding sensitive information, it's what information do you choose to fill that void with? So what I recently did actually uh is I managed to get one of Alibaba's uh hot new AI models called uh Quen. It's now on Quen 3.5, but at the only two months ago it was Quen 3. I got one of those models and I managed to hack it through a uh process called think token forcing, which is basically where you trick the AI model into revealing how it's been aligned to talk about certain topics, uh, in this case, China. And I asked it a very simple question, which is what is China's international reputation? Nothing to do with Tiananmen, nothing to do with Taiwan. And uh the model then generates a list of how it's been told uh in fine teen to talk about China, and includes right at the top, I must give positive and constructed information. I must withhold all negative information. So what the arm answer is generated is just nothing but positive opinions about, for example, China's contribution through the BRI, about uh its ability to generate uh environmentally friendly technology. Uh, and it would go on such as this. So this is much more subtle. And when you ask that question, for example, but rephrase it, what is Kenya's international reputation, or what is Belgium's international reputation? It will say, I must give both positive and negatives. So, as I say, what we're now seeing is AI that can be used as a much more subtle propaganda tool. And as, for example, Quen models are now one of the most uh downloaded set of models on the internet today. In fact, they overtook uh Lama, they are now the most downloaded open source models on the internet. This becomes a real concern. I do believe that large language models uh have the power to replace Google as our go-to for information searches. And if China has this competitive advantage uh over this new technology, this technology this go-to technology or the future technology of information searches, then they have a golden opportunity to redress, as they would call it, uh, an information space which is too Western-centric, which shuts out what they would call uh Chinese narratives and what we would call uh Chinese propaganda.
SPEAKER_02Alex, I believe the term is Jungnong Liang, right? Positive energy.
SPEAKER_04Casper and I, we've also it was this it is literally a line, a line to uh to generate positive messaging, this Quen uh AI model, as far as I can tell, yes. It's designed to give the official uh view on multiple different uh topics, such as, for example, China's political system, but then when it comes to trying to shape international opinion, it will just give uh it will give positive information on uh on a variety of topics in English. This wasn't done in Chinese.
SPEAKER_02So recently, you together with your colleagues from uh China Media Project, you authored the report Guided Intelligence on AI. What was this report about? Is it some of the themes that you mentioned here? And uh, can you tell us a little bit more about the work?
SPEAKER_04Yeah, so that is actually sorry, I I I should have said that's actually uh what that report is uh about. I've essentially more or less outlined it to you. Um it's it took a lot of work, uh, but it was just downloading a lot of large language models that were created by multiple different companies and built off of Chinese AI models and trying to see how effectively they were able to get uh certain Chinese public opinion guidance or positive messaging out of it. And so I tested across 10 different models at least. Um, and I was finding uh this form of information guidance prevalent in all of them in some form. Some had trained it out better than others, uh, but it seemed like it because they had built off of this base model, it was impossible for them to remove it entirely. And it would come up in very erratic and unpredictable ways. There was one model that came out of Malaysia where all it took was an accidental typo in one uh question to go from a relatively neutral international uh answer oriented to international information systems to uh essentially the Ministry of Foreign Affairs spokesperson speaking to you from through the machine using classic uh government uh government slogans. So yeah, that's uh that's what we found. But it was also just a broader investigation of China's uh international AI strategy and how it intersects with uh yeah, with uh their attempt to, as I mentioned earlier, to what they would consider re-arranging a supposedly, in their eyes, unfair information space internationally, which is dominated by uh Western legacy media and emphasizes uh what they would consider unfair or what they would argue are fake stories, which China would like to redress with its own narratives.
SPEAKER_03So, Alex, um how do you see the Chinese approach to AI differing from that of Western countries, both European and American uh as well?
SPEAKER_04Yeah, there are there are multiple there are multiple ways they differ. Um firstly, because of course China has the handicap of not having access to the most cutting-edge AI chips. Um this is because of uh the United States export controls, which have recently been loosened under the Trump administration. So a lot of their strategy at the moment is trying to find ways around that strategic bottleneck. Uh and so we're finding interesting uh papers coming out of uh Chinese AI labs that are addressing efficiency problems with AI models. So AI at the moment, across the board, everywhere in the world, has an efficiency problem. AI models run on humongous amounts of electricity and are still not they are, how to put this, they are almost on a parallel now with human intelligence. And yet at the same time, human intelligence runs on whatever you yourself had for lunch today, which is much more efficient in terms of power. This is something that if uh the current way that uh the US AI strategy has gone so far is inevitably going to have to reckon with. Um but because of the export controls, they I would argue they've given China a head start in terms of working out how to uh address these efficiency problems. Uh so as I say, that's one part of their strategy, which uh the US doesn't really have to deal with because it has all of the resources. And then secondly, they're not really looking to do what the US is, uh, such as uh anthropic, open AI, the rush towards AGI. You have uh Chinese AI labs saying that their raison d'etre is also to create AGI, but the government who is more uh interested in trying to create uh technology, supposedly, uh that can help ordinary people's lives and that can be solved. Um, you also have much uh stricter national planning from the Chinese side than you do in the US. As I mentioned earlier, the State Council has this desire to implement, to push 70, uh, to put have an AI penetration rate of 70% uh into the country as to whether or not they can do that uh is uh up for debate. There are some who believe that they can't. Uh but in terms of that, but then you see uh the United States trying, or rather the federal government trying very hard to ensure that there is no uh states-based regulation, but then again also repealing uh the Biden executive order. The Biden executive order, I think it was towards the end of the Biden administration, President Biden implemented an executive order where which would allow for from greater scrutiny of ch of of AI labs and their uh their safety testing. Um, the Trump administration, I think it was on the only the second day of his uh presidency, he repealed that executive order. So it's only in a handful of states where you have safety regulations being imposed uh on these multiple uh AI labs, whereas China has much more of an emphasis on the regulation side. And I suppose because they are trying to penetrate AI into multiple different parts of society, I suppose you could argue, and indeed some people do, that this emphasis on balancing development with safety in the long run is going to be more effective than what seems to be the US policy, the administration's policy of laissez-faire. Finally, the US is uh quite isolationist when it comes to international AI governance, whereas China is trying to expand into this space uh as fast as it can, trying to encourage multiple different countries to create up frameworks uh whereby they can uh share or rather come to an agreement on how to regulate AI. And also, as I mentioned earlier, trying to set up uh an international mechanism to improve AI governance. So as uh the US retreats in international AI governance, uh the Chinese government is trying to expand internationally. So I hope that wasn't too long-winded.
SPEAKER_03But there we are. I mean, we're called the China nerds, so there's definitely room to just geek out on whatever you feel you want to geek out on. Um what what are what are you touched a little bit uh upon it, but what are some of the real implications of China's AI development for Europe, also with AI governance and the global development initiative? Because we've talked in the podcast a lot about Xi Jinping's many initiatives, many global initiatives. Um on the AI side, do you think that Europe is even aware of where China's headed and how fast they're uh they're going there? Oh dear. Um well I'm I'm a big question, I know.
SPEAKER_04I'm afraid I'm I'm afraid I'm based in Taiwan, so I I I I feel like you might know have a better answer for this than I do. Um but I suppose you could say I get the sense that there is uh already raised awareness amongst the general population uh within the EU that uh, well, possibly besides Hungary, um, that there is uh that that Chinese technology has certain uh uh guidances or certain values encoded into it, which may differ from uh what are European values. Um I think the problem is going to be uh whether or not those that awareness, that lack of trust actually balances out, for example, in an AI startup, uh the costs. Um so as I mentioned earlier, Chinese AI is predominantly, not entirely, but predominantly open source, meaning that uh multiple uh AI startups can already save a certain amount of money by choosing an open source form of the technology. Uh as to whether or not they trust it on certain answers, uh, they may actually be more concerned about the bottom line of their, I suppose, their margins. So yeah, I think they're much more uh that's the way that I think it's more likely to be adopted. Especially as if I remember rightly, the EU AI Act stipulates only that uh someone has to be told that they are being spoken to by AI, not which AI model is being used, though. So I think that may be a way that uh AI becomes adopted, uh Chinese AI becomes adopted in the EU, whether whether or not uh there is a general sense of trust of it.
SPEAKER_02Alex, I wonder with your uh uh insight and also all the research you have done in in AI and the focus that we now see from different parts of the world to actually adapt uh values to different continents, different um you can say centers of power. Do you think we're moving towards uh you know American tech uh and AI companies having a certain uh uh kind of value set that are um propelled from there, EU having another one, China having a a third, and so on. Because AI and the different ways you can uh you can influence narrative across the world is simply so effective. I mean, you are much more in uh in the nitty-gritty in this field than than I am. So sometimes when I see both the Chinese strategy on this area, but also how we discuss this in Europe now. There's a big discussion in Europe about building up our own tech companies, uh strengthening our own AI tools, perhaps to break away from the ri reliance on American uh uh tech. I know it's a it's a broad question, but I think it is really an important question of our times and perhaps a uh a direction that we're seeing, but I would like to hear what you think. Sure.
SPEAKER_04So I think every single government around the world has suddenly realized that AI is now not just about AI, it's about power. Um I think the EU is aware of that. I the US is certainly aware of that. The US AI Action Plan says that the US needs to ensure that it quote unquote dominates AI going forward in the international space. Um, Xi Jinping has called AI having the potential to have a quote-unquote lead goose effect, which I I just quite like as a term. And what that means is it's it's it's in reference to a flock of geese in flight in a V formation. The goose at the front is the one which is dictating the direction that all the rest of the other geese go. Um, so I think multiple different power centers, the EU, the US, China, all around the world, know uh that this can be a way to project their own values out into the world through harnessing and leading uh this new technological revolution. You also have multiple different governments around the world in the global south realizing that this is also their chance to uh influence regional politics, uh launching AI action plans of their own in all sorts of different parts of the world: East Africa, West Africa, South America. Um so I would argue actually, in that way, in that sense, China's uh desire to harness the power of AI is actually no different from any other international government around the world.
SPEAKER_03Maybe we can touch a little bit more about uh the role of media analysis in understanding China today. How would you describe uh the role and and the importance? Uh the th sorry, the role of media analysis in uh sorry, in understanding in understanding China today.
SPEAKER_04The role of media analysis in understanding China today. Yeah, so uh I think I've kind I've I've briefly touched upon it. Um The problem with being external or outside of China at the moment is that unless you have uh contacts, personal contacts within China, and you're not on the ground yourself, all we really have to go on is media from China, be it social media, uh traditional media, uh AI-generated media. It's all more or less the same thing, and all of it is, as I said earlier, part of a system uh which is tightly guided by the party. It's about curation, it's about presentation, both domestically and internationally. So trying to kind of penetrate that, I think is still very important. Uh but nonetheless, when it comes to media analysis, as I mentioned earlier with my conversation with this uh taxi driver, it's still very important, I think, to know how the majority of uh the mainstream audience in China are going to think about a certain topic because of the way that the information space is being guided by government. I think you do another example uh where during COVID, uh I was sitting in uh a restaurant and struck up a conversation with uh one gentleman who started telling me, oh, well, it's I feel so sorry for you, uh you democratic countries. Uh because you have this democratic system, it means that you can't contain COVID because you can do whatever you want. Whereas in a one-party system, we have been able to eradicate uh COVID. This was in 2020 through to 2021, when admittedly China had done a very good job. But at the same time, so had South Korea, so had Taiwan, so and Japan, all democratic countries. So this was quite clearly something that the party was uh using this success to this uh they were using this success in order to bolster their own uh credentials among the populace to prolong the social contract, whereby we uh you have no democratic freedoms, uh, but we still keep you safe nonetheless and look after you.
SPEAKER_02Alex, I would like to ask you a little bit more uh regarding your time when you were writing from China, for example, when you were writing for uh the China Project and also for the world of Chinese, because Casper has been uh posted as uh a correspondent, but uh but you were but you were there as um uh I I guess first freelance uh writer for uh China Project and later as a um employee at the China Project. I sorry at um the world of Chinese. How did you find it to um to to write these stories in China? Um I remember you and I uh we um we were in the Hutongs at one point where you were writing one story about Hu Xi Jin, wasn't it? Oh yeah. I think for uh for the China project. Yeah.
SPEAKER_04Yes, and we um we started talking to that that man at that uh newspaper, at that newspaper outlet, and he started asking why are you asking so many questions? Are you a spine or something? Half joking, half not. Um but it's sorry, just uh I could could you just kind of repeat the question?
SPEAKER_02Yeah, just I think we know several journalists who uh who work as posted correspondents to China, but you were working um as a freelancer writing uh from China, and I guess that's also that that's just another type of uh of experience that I think many of our listeners um they might be interested to know how do you do that.
SPEAKER_04Ah, right. Okay, so this is a problem. There was someone who approached me on LinkedIn uh I think in the middle of 2024, um, and he said to me, uh, I am uh a Yen Ting scholar based in Beijing, uh, and I would like to know, I would like to become a journalist. I would like to know uh how you manage to do what you uh do what you did, and maybe uh that's a career path I can try. And I had to think for a while, and I suddenly realized that even two years later, the career path that I had made to work for me in China is no longer an option. Um, so for example, you have the fact that the China Project, where a lot of uh uh China watchers of our generation found their feet uh and managed to start writing, the China Project has closed down. Um the world of Chinese uh has uh its own difficulties. Uh and then there's also yeah, there's also issues with just, I suppose if you try and write something uh and you are not on the right visa to be asking questions, uh that can be dangerous for you. Uh you could do what some uh people do, which is go on a student visa and then try and gather information and publish it as soon as you're outside China. But that also can be a rather risky strategy because what if it's something uh that is uh dangerous to say? Uh there was one person who wrote something for the FT, uh, who was a Yen Ting scholar, and he now is has been advised not to go back to China by his uh former professors. So it's it's a bit of a tough situation. Uh, and people I speak to in this space who are journalists still based in China, uh, they find it uh very difficult. Um, for a start, if you're a journalist as a US-based outlet now in China, it's highly likely that you're not going to be able to get uh a J visa, that is a journalist visa processed to go into China. Uh and I believe that there has only been one J visa processed uh for a US uh outlet uh sin for the past four years, I think. And that's because uh I think there's there's supposedly a mate uh a media going on uh where the Chinese leadership is unhappy about the way that uh the United States has uh designated uh certain Chinese uh state media outlets, and so this is meant to be some form of war of attrition, at least from what I've been hearing from journalists inside China. So you're not able to get journalists coming in to replace former US journalists. So I suppose your option would be to try and do some form of writing for a European paper, perhaps. But even then, as I say, if you're freelancing there is continually that risk. I would say that the work that I was doing when I was freelancing was relatively safe. I wouldn't say it was uh something to do with Xi Jinping. I tried to steer clear of politically censored topics because I didn't want to be thrown out of the country. Um yeah.
SPEAKER_03I mean now you're you're as as a lot of people are trying to gauge China from from abroad. Obviously, I mean the lack of of information and the lack of of people is something that is is a big issue when you're you're doing that. And for Denmark, we we only have one correspondent based in um in Shanghai, and then one based in in Hong Kong who can access mainland. But other than that, I mean there's there's one person who's covering all of Asia, not just uh China. But what do you see as some of the biggest challenges despite the obvious or aside the obvious of not getting access to talking to average people? Um and also some time ago, Mass and I interviewed um Emily Feng, um the journalist of uh National Public Radio in the US, and she said some of her best stories just came like randomly appeared just by talking to people. So just the lack of access to talking to your Lao Paising or your average Chinese is of course a big one. What do you see as some of the big hurdles?
SPEAKER_04Well, I think the most important one is that you lack nuance, I suppose, in your coverage. And that's just an unavoidable problem, I would say, at the moment. If you're not on the ground, you can't uh see, for example, past uh what the information feeds would like you to see. One example I can think of is uh during uh the zero COVID lockdowns in Beijing in 2022. Uh there was one article from a major international newspaper uh which talked about the difficulties people in Beijing were facing. They were unable to go to restaurants, they were unable to go here, there. But actually, as I was on the ground, you'd just walk along the street and you'd see everyone trying to exploit every single loophole they could. Yes, rest people were banned from eating in restaurants, so what they would do is just take the tables outside and do da pie da, like just just eating out, as it's called. And so, yeah, because this journalist wasn't on the ground, she couldn't see that actually it was a lot more uh varied, and people were finding ways around these uh these these regulations. I would say that's the most important thing. And this is a tragedy of the way that the system has been set up, uh, is that these uh the journalists can no longer go on the ground. They can no longer get that nuance. And indeed, it sometimes even feeds into uh things that I see from the Global Times saying that Western journalists don't understand China. You see, their reporting is so unnuanced. But that's of course because these people are not allowed to go on the ground. So yeah, that becomes a weapon in and of itself that is uh used by this by the state media apparatus.
Projects for the future
SPEAKER_03I mean, we're we're of course all biased here because we're we're China watchers, we've been embedded with China, we're dealing with China for a long time. But can you think of any other country in the world where we have such a big information gap um relative to the importance of that country compared to China?
SPEAKER_04Okay. Um this is not going to be a popular opinion in the United Kingdom, but there are certain ways in which Buckingham Palace and Jongnanhai, their information flows are very similar. Um here are the similarities. They are more or less a black box of information, but there is a huge amount of demand for information to come out of it, meaning that even the smallest thing that is seen from either Buckingham Palace or John Lanhai becomes something that is blown out of all proportion uh or misinterpreted. Um, and it's opens the floodgates for all sorts of rumours. So, for example, you have Xi Jinping, if he disappears for one or two months or three months every summer, suddenly there is frenzied speculation, oftentimes driven by uh Fallon Gong associated uh YouTube accounts saying that he's been assassinated or there's been a coup. Um and then if Cade Middleton suddenly disappears uh for a certain amount of time, uh people start assuming all sorts of different things and peddling conspiracy theories as well. Um so I'd say that's just a general trend. If there's a black box of information and yet there is huge demand for it, and there is uh not much uh information forthcoming to satisfy that, then it does lead to uh to the same effect, I would say. Um yeah, does that uh does that answer your question? Yeah.
SPEAKER_02Alex, now since you returned to uh to Taiwan from a recent uh trip to Europe, what are you working on now? If you uh can share that with our listeners, what are some of the projects that that that you're looking at uh for the foreseeable future?
SPEAKER_04For the foreseeable future, it's going to be monitoring uh the expansion of Chinese AI models into different parts of the world uh and China's uh attitudes towards regulation, towards AI safety, and how that is going to feed into uh its long-term geopolitical goals, including uh, yes, I suppose in terms of uh, for example, it encouraging the uh adoption of, for example, its non-interference policy in terms of uh geopolitics. Um because this is something uh that we are noticing. Uh AI is developing very fast as well. I would like to do more work now on AI agents, given there has been a sudden appearance in the mainstream, both in uh Europe, America, and China, of uh AI agents that can be used by consumers uh problematically and with very little safety regulation, but nonetheless uh they can still be used. Uh how is the party going to regulate that? Uh genuinely there are safety issues uh with the way that AI agents currently work, but are they going to be uh shaped in accordance with Ren Gong Junang and Xia, uh AI safety? Uh China has something known as the comprehensive national security concept, which has different strands coming off of it, uh, which includes uh I think ecological security, uh information security, but then also AI security now. That's a relatively new addition. So, yes, uh how are AI agents going to be made safe by uh Chinese staffers?
SPEAKER_02So earlier Casper asked what the implications are um for some of these strategies relating to AI, they are for Europe. And I would like to ask that now, since I think we're seeing a more assertive sort of thought process from uh from Europe and different leaders in Europe on uh on on trying to fret out uh a uh European direction on this theme. What do you think we can, I don't know, uh uh uh learn or how some of uh looking at China's uh approach to AI can inform what we can do in uh in Europe. Do you understand the question or does it make uh make sense? Oh kind of?
SPEAKER_04Yeah. So uh the way that uh China or at least Chinese regulators look at the way that the EU has tried to govern AI, um they say that uh the EU's AI Act actually smothers innovation. It means that uh technology companies are too bulked down with regulation to be able to actually move as fast as they can in this space. What the Chinese side does is emphasize standards, which are non-binding, uh collectively shaped by uh different uh players within the AI industry, but under the party's remit. So uh I think what we're seeing is that the EU AI Act is being uh streamlined. So uh perhaps the EU has already taken that into account. Um I think that's the best I can give you in terms of an answer on that one, I'm afraid.
Outro
SPEAKER_02Sure. Alex, uh I think we've covered a lot of uh of uh great angles on the um on the subject. And I was thinking that if any of our listeners are interested in exploring uh the theme of of AI and China, but could you recommend some different resources or maybe books? There are always already a number of books out there, but uh I would actually say probably too many on the subject of AI and China. So what do you uh find uh some like what do you you find as good resources that you could recommend to to people who are perhaps not as acquainted with the theme?
SPEAKER_04Okay. Um the problem that you have uh is this information space, or at least AI, is now moving so fast that by the time someone writes a report or by the time someone writes a book, it's already kind of out of date. So I wouldn't really recommend a book, per se, unless it's The Empire of AI by Karen Howe, which gives you just a great grounding in how AI as it currently stands works and how it was uh shaped in Silicon Valley. Because most uh Chinese tech workers in AI labs are still watching what happens in the US very closely, uh, and are most likely to adopt their next strategy or idea based on what America is doing. So I still think that would be relevant. Um Frankly, it's just a lot of newsletters that I have. China Talk does work on AI, so does uh Jeffrey Ding, who runs the Chin AI Substack. Uh there's also, oh, let me see. There's Interconnects AI, which is uh run by a man called Nathan Lambert, which is also on Substack. Uh there are multiple different uh players in that field who are doing good work. Um and yes, I'm afraid to say I'm afraid to say also Bill Bishop. In fact, no, I'm not afraid to say, um Bill Bishop's uh cynicism, uh, which was my education when I started out uh in China. Just reading, I subscribe to that newsletter uh and I still read it religiously every day.
SPEAKER_03I mean, he is the guy for a reason, and he's been doing it for such a long time. So I agree with you. There's no shame in saying uh Bill Bishop. I think the man uh maybe his dues as well.
SPEAKER_04No, but that that newsletter is very essential, I'd say, if you were just starting out. Uh in fact, it's essential even uh even now.
SPEAKER_03Do you have anything uh concluding for our listeners about China, Taiwan or in general before we wrap this up?
SPEAKER_04No, not off the top of my head. I'm sure something will come to me as soon as we uh as soon as we stop, but not off the top.
SPEAKER_03And and also for our listeners, if people want to follow your great work, both for China Media Project and other reports, what's the best platform? Is it x.com, LinkedIn, or where can people uh find you?
SPEAKER_04Yeah, we have a substack uh which is called Lingua Sineca. Um uh that has multiple different bulletins on it. I write the China chatbot substack, which normally comes out twice a week, uh, which is just keeping up with the latest uh uh developments in the Chinese AI ecosystem policy regulation frameworks, all of that. Um and also China Media Project. I I write articles on there uh whenever yes, whenever time allows.
SPEAKER_03One one final question now that you um Yeah.
SPEAKER_04Just I just thought. Um there's there's also I really would recommend um the work in terms of if you want to know how China's AI policy is working, uh Scott Singer and Matt Sheehan over at the uh the Carnegie Endowment for International Peace, they also do very, very thorough research. Um yeah, sorry, that was another thing that just came to mind.
SPEAKER_03One final question now that you reminded me about it. The the Lingua Cynica, could you briefly explain what what that is and why people should follow it?
SPEAKER_04Arts, uh heart. So are you talking about Lingua Sinica the website, or are you talking about Linguistineica the Substack?
SPEAKER_03Um the Lingu Cineca that you work with in um in China Media Project.
SPEAKER_04Uh yes. So there's Linguist Cineca, which is uh run by my colleague Daria Perete, which is a monitor uh for multiple different uh media engagements by uh Chinese media companies around the world with different with other newspaper companies or traditional newspaper companies, traditional newspaper outlets, uh just in multiple different countries around the world, including uh reports on the media ecosystem on the ground and how much influence uh China currently has over it. Um it always amazes me what that part of the team finds because essentially because there really is so much once you start going looking.
SPEAKER_03Alex Covell, um I think we've come to the end of the program. Um on behalf of both myself and Mass, thank you so much for for taking some time to be our guest here on the podcast. No, thank you very much.
SPEAKER_04Thank you so much for having me.
SPEAKER_03And to all our listeners who do follow uh all the substacks that uh Alex recommended, and especially China Media Project. I think it's uh it's a great resource that not necessarily a lot of of Danish people know about and should utilize and use more. I mean, it's a great insight into a lot of aspects in China that helps us get China right. And as you all know, one of the main themes of Mass and I is that it's never been more important to get China right. So thank you again, Alex, for being our guest.