AI’s Impact on the Chinese Economy

Show video

Greetings, everyone, and a warm welcome to the second episode of our webinar series on US-China competition. Today's event is hosted by Asia Society Policy Institute's Center for Center for China Analysis. Thank you so much for joining us as we discuss this super important topic about as far reaching impact on China's economy. Today, we've really got a stacked agenda that will give you a comprehensive understanding of how shapes different sectors in China, challenges, as well as opportunities that lie ahead, and the ethical and data security considerations that come with exponential growth.

But first, let me introduce our panel of experts who will be guiding us through today's conversation. First up, we have Mr. Nigel Inkster, director of geopolitical and intelligence analysis at Enodo Economics. Mr. Inkster is someone who has over 40 years of experience, a wealth of knowledge, particularly about China. Welcome, Nigel. Thank you very much. Next up we have Caiwei Chen, a top tier china reporter at rest of world. She will offer us that critical journalistic lens through which to view ai and its interaction with china's economy. Thank you so much for being here. Thank you, Lizzi, for having me.

Our third panelist is Mr. Michael Frank, a senior fellow at the Center for AI and the Advanced Technology at CSIS. Michael has spent a decade in Asia and is currently focusing on the meeting point of geopolitics in technology. Thank you so much, Michael. Thanks, Lizzi. Last but certainly not the least, we have Ms. Elsa Kania, an adjunct senior fellow with the Technology and National Security Program at the Center for New American Security. Elsa's research orbits around Chinese military strategy, military innovation and emerging technologies.

Thank you. Looking forward to the conversation. So incredible lineup of panelists. And we are going to dive deep into the subject today. Let's get started. So I'm going to turn to Nigel first. China has made it its national security strategy to be a global leader in AI Nigel, can you talk a little bit about how AI fits into China's national development priorities? Indeed. Well, I think it goes back a long way. China, as we know, was caught out by the Industrial Revolution. It missed it.

And, the history of modern China has been a history of catching up with and surpassing the rest of the world in advanced technologies. China came late to the party when it came to information communications technologies, but it caught up with impressive speed. That said, it is still very dependent upon us and other Western technology. And I think it wants and it's quite clear, they made it very clear in numerous policy statements that their aim is to become a world leader in the next generation of advanced technologies.

And of course, this includes artificial intelligence. And there are several reasons why this makes sense for China to want to do this. Firstly, at the moment they're paying Western companies, Western technology companies large sums of money, importing semiconductors, paying royalties, etc. But I think China would like to be in the driving seat when it comes to the next generation and collecting royalties rather than paying them out.

The second thing is that for China, domination in these technologies is something that confers significant geopolitical influence, the ability to shape events that we've now seen. Just today, China issued a white paper, in effect proposing a dramatic remodeling of the post-World War Two US led international order. So domination of technology will help with that. And thirdly, and this is an insight I owe to my Enodo colleague Ms. Choyleva. I think that China looks at a technology such as artificial intelligence and thinks that this can be used to deliver a form of authoritarianism that is more precise, less broad brush, more scalpel than than a sledgehammer to, to to to control those aspects of societal behavior where whether social financial than they're currently able to do.

there are many other areas where China sees AI as conferring significant advantages. The military dimension I won't talk about that because Elsa has incomparably better understanding of that than I do. But it really is absolutely central to everything China wants to do in its efforts to achieve its two centennial goals. Thank you so much. Michael, how is AI contributing to China's geopolitical ambition in general. Yeah, I think you can break down that contribution into two parts. The first is the broad economic base that you need to realize those ambitions. And then the second part is the actual power projection capability.

But in terms of that first part economics, China has a shrinking labor force that is relatively unproductive compared to its ambitions for where it wants to be in global value chains. So China's GDP is 75% of the US's, but productivity is just 27% of U.S. productivity. If you're trying to emulate a US style consumption led economy, obviously this is a this is a huge problem. Now the good news is, I disproportionately benefits high value industries so China can move up the value chain using AI, but it also has to. in terms of creating a new tax base that moves out of this property led development model where local governments are raising tax revenues through through land sales and then reinvesting them in infrastructure projects, that is coming to an end.

So it's very high stakes to make this transition. It's probably the only possible solution to make this transition when you have, such such a dire demographic environment. So to be able to grow income and increase productivity at the same time. There's not a lot of evidence that this is happening yet.

So if you look from 2012 to 2019, China's productivity growth is about 6% a year. And over the last three years that's slowed to closer to 2%. That's obviously still a lot higher than advanced economies, but we're talking about them moving up the value chain. So it's not just you relative competition. They really need to to boost productivity growth to move up that value chain.

And if they don't, I mean, I think the precedent is pretty horrifying from an economic perspective, if you look at Japan. So in terms of, the actual power projection part of this, it's really difficult to measure. But surely, if you look at what the US Department of Defense is doing with the Replicator initiative, I'm sure Elsa can can speak to that better than I could. Surely China is doing a version of that with the play just with less openness than we have in our system. we put a spotlight on our problems which make it look like we're not making that transition to an AI driven national security and geopolitical strategy.

But that also creates a political pressure to carry out painful reforms. And I think that has happened in the US over the last few years. comparatively, China's still working on rooting out corruption in its military. You look at the recent detention of Defense Minister Li Shangfu I think that's a good example of where the types of problems that the PLA is dealing with. there are some pretty big differences there. Having said that, there are some big similarities, too. it's it's common, a common talking point in Washington to talk about how nefarious China's civil military fusion is.

The U.S. has the exact same thing with AI. If you look at all of our leading AI companies, they're all participating in projects with the U.S. Department of Defense. The whole point about AI, why it's so sensitive is it is dual use. You can't neglect either the civil applications or the military applications.

So my suggestion would be if you look at how they are reaching their how they're going to be able to effect their geopolitical ambitions, You look at comparable Chinese companies, see where they are. Technological baseline is they're behind at the moment. But keep in mind, they probably don't actually need world leading AI to realize those geopolitical ambitions. If you have any kinetic conflict between the US and China in the next few years, it will probably take place in Asia, where China has enormous home field advantages. And there will there even with sophisticated U.S. technology, that's that's not going to necessarily be the decisive factor when you have all of these other environmental advantages that that China certainly enjoys in that theater.

And thank you so much. Elsa, what strategies is China currently employing to transform different industries? We know we talk a lot about transforming different industries, but how is China making this transformation a reality? Where do you see having the most substantial economic impact within China's various industrial sectors? As my fellow panelists have alluded to, we've seen the Chinese leadership highlighting the importance of artificial intelligence at the highest levels, especially when it comes to the need for China to achieve this economic transformation and to sustain growth. The idea of innovation driven development, I think, is very much reflects the notion that technology is really critical to China's transformation, to leveraging emerging industries to compensate for slowing growth. And we've seen this coming into focus since the 2017 plan that has really highlighted the importance of combining the dynamism of China's commercial enterprises that have already become leaders in AI in their own right with robust, robust state support and highlighting a number of emerging industries. Of course, we've seen the digital economy and platform economy economy be an emphasis and a lot of apps and a lot of applications were.

AI is being in some respects almost seamlessly integrated in today and day to day life in ways that that are sometimes powerful but provide a lot of convenience without necessarily being truly disruptive or transformative necessarily. There's always the joke that once AI is successfully implemented, we start to take it for granted and it almost becomes invisible, whether that's a navigation recommendation, algorithms and the like. Of course, self-driving vehicles have been a area of the industry where China's really leaned in and we've seen a favorable environment in terms of policies and support and a lot of companies coming into the fray and I think a lot of potential there going forward.

So that definitely will be an area to watch, of course. Yeah, a lot of interest in cross discipline intersections, the applications of artificial intelligence in biotechnology and health care, of quantum information science as well. And I think going forward with the demographic challenges as have been discussed, the relevance of intelligent manufacturing is apparent to robotics automation.

And in a factory context, to compensate for the shortfalls in the workforce. That is certainly something that is generating a lot of enthusiasm and moving up the value chain through leveraging robotics, sometimes enabled by 5G and a focus on sort of incorporating these advanced industries really into the heart of China's new model for development in a new era. Xi Jinping has has declared it. So I think whether it is future of health care or the future of logistics and transportation or the consumer economy, definitely a lot of scope for interesting and sometimes sometimes surprising or sometimes more or more routine day to day applications of artificial intelligence at this point. Thank you so much, Elsa. So Caiwei, I'm going to turn to you and can you give us some examples of possible or case studies where AI has significantly changed a certain industry or sector in China? Or do you see more challenges as China rapidly integrates technologies? Certainly I would love to.

At Rest of World, me and my colleagues has been focusing on e-commerce ecosystem in China. We spent a lot of time talking to the vendors, the sellers, the live streamers that are selling the product and the buyers. I really want to highlight the global ecosystem, which is China's biggest shopping shopping platform. We really see a rapid integration of AI in every segment, every action in the shopping process. We're seeing Taobao, the company just testing out a large language model based chat bot one, which is based on the company's large language model me. And it's an online shopping assistant that's going to help you to make your shopping decisions.

For example, if if a customer wants to buy a camping supplies or you need to do what you used to need to do, might be do some research and make a shopping list. But with the chat bot, the customers will be able to just talk to the shopping assistant. Ask them what you would need for camping, for example. You can even like do back and forth with the chat bot saying, Oh, this, I don't like this dress. I need something more formal.

So any, any kind of conversation or prompts, well, we'll return a mix of content. It will have the links of products and livestreaming links, even blog posts in it. So that's from the customer site and from from the e-commerce seller side. We're also increasingly seeing air being applied in the selling our product to marketing. And we're already interviewing a lot of e-commerce store owners that are using tax to image generators like DALLE or Midjourney to produce their product photos.

So an example would be they can just they can just put like a piece of clothing that they want to sell on a mannequin and then use a program to transfer that into a real, real looking demonstration photo. So that will drastically reduce the cost. Not to mention we're already seeing a lot of e-commerce air life streamers which are not real humans, but a model that can work 24 seven.

So a lot of people say these e-commerce live streamers are less organic. This one is less organic than a human life streamer. But on the flip side, with a lot of the political sensitivity sensitivities, famous Chinese livestream where it's like Li Jiaqi are getting these days. It might be a good sign that these air life streamers are less risky because they only say the pre-written script and perform preprogramed movements. So I think all these applications in the e-commerce system is a great demonstration that air has already been integrated in Chinese economy in a lot of different aspects. Thank you so much. Lots of potentials as Caiwei and Elsa alluded to.

But Michael, what kind of strategic investments in innovations are still required for China to unlock the full potential of AI across those various sectors? Yeah, well, I think of this both in terms of horizontal issues that affect every sector and then the types of verticals that Caiwei was just describing. And starting with the horizontal issues, it's finding a way to match international standards or the sophistication in terms of computing power and capital. I think this is going to be really tricky for China. And actually, if you look at the Stanford AI Index over the last five or six years, even before we've gotten into this new US policy of actively trying to curtail China's AI capabilities, the gap between China and the US in AI is actually not closing, which I think is is not where we thought we would be if we look back five or six years ago.

And as Nigel described, China's history over the last few decades has been closing gaps with international tech sophistication and then and then ultimately surpassing peers. But that doesn't seem to be happening with the AI. I'll talk about talent in a minute because I know we were going to we were going to talk about talent later.

But thinking about computing power, this is going to be really hard. Look at the timeline for how quickly AI becomes obsolete. So last year GPT three was productized in November. It had been around for for a for a while, but it was only productized in November and by March it was already obsolete. So even if you're willing to throw endless money at a problem and there's evidence that Chinese firms are the timeline of working on inferior hardware, really it just is. That is this speed limit that precludes any sort of world leaning AI capability.

I do think that Chinese firms and with the support of the government, will find some clever ways around this. And already you are hearing about some, ingenuity, clustering, computing resources to try to improve performance. But it's going to be an enduring challenge.

But then, you when you think that you get to the vertical piece and some of the things Caiwei was describing, it's it's important to remember that diffusion is not the same thing as cutting edge. And in terms of unlocking AI's full potential, when I look at, the corporate level or the economy level making the most of this technology, it's in diffusion. And when you think about these positive cases of consumer facing Internet applications, China's information and communication technology companies are as sophisticated or more than their international peers when we think about AI broadly. That's already the case now and even some other interesting sectors. I was at Economist Impact before, and we had put out a study called the Intelligent Transformation Index, looking at China's sectors and their progress on AI and some surprising sectors where China's world leading manufacturing is one. You have Chinese manufacturing firms disproportionately use AI compared to the international peers.

And construction, which is one of the is perceived as a lower value add industry. Chinese firms are even in that relatively low tech field, are using AI already. So I think those are areas where you're going to see broad diffusion and it's more likely that you will have China reaching full, full potential there. I think the negative case I do have to call out is, there are some sectors where China really does lag its international peers in terms of tech sophistication now. And I think it's there's some enormous headwinds to that changing.

And the two that stand out are, number one, finance. China's banks really lack sophistication compared to American peers compared to, even think of a bank like DBS based in Singapore. They've got a lot of issues to sort out in terms of digital transformation and financial product specialization before I think they're going to get to the point of really leveraging AI.

And the second area is health care, where again, you have, an aging economy that already, the Chinese health care system probably does not deliver consistently the quality of care that the government would be satisfied with. And it's only going to get harder to deliver that level of care. Maybe AI is the path forward, But, to to borrow a line from earlier is, they really have to use AI because there are no other options. So it's very high stakes. It's an enormously complex ecosystem, but you really need a good foundation of care infrastructure to build on, which China does not have at the moment. Fantastic. Speaking of headwinds, Nigel, the economic landscape in China currently appears increasingly challenging with escalating local debt, aging demographic property market on the brink of some sort of drastic decline or collapse. I mean, the situation really raises some very concerning questions. Do you think China has reached its peak economically or are there still untapped avenues for growth that the country could potentially capitalize on?

Well, firstly, I should emphasize the fact that I'm not an economist. And so you have to take what I say with a certain amount of caution. But I think and I think it's probably true to say that the original Chinese growth model has pretty much run out of road, and that's not a new perception. I think China's own leadership were becoming aware of that pretty much around the time of the 2008 financial crisis. this was not sustainable. And Stein's law kicked in. What you know what can't continue has to stop.

So they're in this situation now where the old model they know is no longer functional. What they need to do urgently is to stimulate domestic growth and consumption. But just at the moment, because all the traditional stores of value property market in particular are hitting real problems And also the aftermath of the COVID lockdowns, it's quite clear that China's population as a whole is behaving very conservatively at the moment, stockpiling savings against a very uncertain future. Particular problem with youth unemployment, which has been well documented or actually lately not well documented because the Chinese government have stopped producing statistics.

But all all of these are factors that are disincentivizing growth added to which, of course, you've got an international situation in which confidence of international investors in the Chinese market is also significantly eroded. But I don't think that that's the end of the story for China at all, that there is a lot that they can do. A lot depends on whether they are willing to take the decisions that will enable this to happen. One of the easy ways to get things moving is to put money in people's pockets. The Chinese Party state has been very reluctant to do that because the region thing has a particular take about welfarism, as he refers to it. But actually doing something about welfare benefits, putting money in people's pockets could have a significant impact. I also think we need to be very careful about drawing long term conclusions from immediate crisis situations.

if we look at Japan, the so-called lost decade of the 1990 is actually life for ordinary Japanese was not that bad during that period. China is in a different situation for the reasons that have been mentioned are much lower levels of productivity. And also still, they are navigating the perils of the middle income trap. But I certainly don't think, China is run out of road yet. And as has been mentioned, that the ingenuity of its entrepreneurs and engineers is such that, given its head, I think they will be able to find ways to generate a new pattern of growth. Fantastic. Caiwei, I'm going to turn to you next as we look at workforce evolution. Talking about employment, a major concern, at least here in the U.S., is job loss due to automation.

How much concern is here among the general public and in China, based on what you hear is high weight? Are there any notable initiatives in China by the Chinese government focusing on reskilling the Chinese workers for a AI driven economy in the future? Yes, Lizzi, this is the job concert is is something that I've encountered frequently in my reporting to apprehension regarding automation. And AI's potential on job security is very, very palpable, especially among younger demographics in China. A lot of them are very anxious and engaged in what we would call semi creative professions that we're observing roles like copywriters, legal assistance, graphic designers, illustrators and even our customer service. These are essentially white collar jobs that have traditionally been valued for the human touch they bring. But now that little bit of human touch is AI is able to replace that.

That's why these jobs are on the forefront to a lot of change that they're going to undergo. But in theory, these positions might be vulnerable. But we must also recognize that it does take considerable time for for the companies to adopt and integrate these emerging technologies.

It takes a lot of time for for the company to offer a mature B2B solution before they actually replace these workers. And there is a waiting period involved. So increasingly, instead of merely waiting in the trepidation, a lot of professionals in these sectors are proactively embracing these technologies themselves, because I do think the development of AI is essentially a tool for for these professionals to use.

And there are vibrant discussions and exchanges on social platforms on how to effectively use tools like chatbots and the Chinese equivalents. So these tools can be seen as AI's rather than replacements, as like a human layer of discretion, judgment and context. Awareness is still very it is irreplaceable. Yeah. So in terms of governmental initiatives for jobs, we are seeing a lot of regulation.

And I do believe in this aspect China is ahead of the US in regulating AI and pulling resources from the government to support youth unemployment problems. And the aspect that often get overlooked is the backbone of a lot of AI models, including ChatGPT rest on human work. We all know that ChatGPT relies heavily on data annotators in Kenya that the company hires.

And similarly, China's AI industry and the large language models have given birth to a myriad of data data annotator positions. So I recently did reporting in how the local and provincial governments in rural China are actually spending a lot of money paying the big Internet companies like Baidu, Alibaba, Bytedance for them to establish data annotation centers in a way that boost like and boost employment create jobs in their local economy. Although these jobs can be very grueling, very repetitive, they do become a sector that local government tend to value. And those governments with resources became a very important lifeline for these tech companies as well. When we see investment, especially foreign investment, is less obtainable. Yeah. Fascinating to see how those, the problems of employment, labor rights and the ethical dimensions of AI intertwine.

And thank you so much for that deep dive, which I thought was super insightful. Michael, I'm going to turn to you next. You mentioned this talent gap in this ecosystem. Can you elaborate a little more? And how has the Chinese government been making efforts to get that gap filled? Yeah, I think there's a lot of nuance here because I think to start, China has lots of elite AI talent. And, in both in in quantity and quality, China's certainly not lacking for for talent.

I think it's a matter of how that talent lines up with with actual commercialization diffusion. And I think it's also important not to overstate the quality of China's AI talent either as being this inevitably world leading force. It is very, very good. But if you look at the Stanford index report, some statistics they have on talent over the last few years, the conversion of degrees to jobs is actually not that great in China. It's pretty far behind quite a few countries that we don't necessarily think of as being on the cutting edge of air diffusion. And I know a few years ago in China, we would always talk about degrees as evidence of China's inevitability.

But I think actually what we're seeing now is maybe it's not a really good, good predictor if the U.S. ever got out of its own way on immigration. I think not only would China be left behind in terms of air talent, but that could actually be a zero sum proposition as Chinese researchers might prefer to study in the U.S., where they have maybe more freedom to research, they have better access to computing power and financial resources. I think that's that's a hypothetical. The reality is for now, the U.S. is not getting out of its own way. So, it'll be interesting to see how that develops.

The China definitely has lots of great AI researchers, and I think both Elsa and Caiwei mentioned, the commercial people that you need to to diffuse those innovations. And Nigel, references to ingenuity as well, of the entrepreneur class. The talent problem to me I think is is much bigger picture. how do you deal with a generation of historically high youth unemployment and underemployment? Because that's not just a supply side drag in terms of skills misalignment for AI, though it's obvious that is happening, but it's also a demand side drag because part of the reason AI is going to be so transformative if you have a broad diffusion, is because it's also going to bring up wages for people and it's going to create this virtuous consumption cycle.

So on the product side, I think that's pretty well handled that in terms of there are there are these great business leaders who have this experience, they have skills in terms of, designing good products that people want to use From, past wave of social media, e-commerce payments, technological change that will translate very, very well. But, we're talking about a pretty big environmental challenge here. When you have over 20% of youth unemployment or underemployment, that goes beyond just what are the capabilities of your your best and brightest.

I also think and this is really, a working thesis, so I welcome any feedback, any anyone has on this. But in terms of, restrictions on international collaboration as a very international field, I think most of them come from the side of the Chinese government. At the moment, yes, the US is investigating the thousand talents, programs and commercial and industrial espionage that's sponsored by the state. But I actually think there's a far greater impact from the Chinese government's general chill of working or collaborating with foreigners, and that applies to all aspects of society. But certainly an area like AI, which is has been clearly marked out as strategically important.

So, the question I would have and if anyone has evidence to the contrary, I would love to would love to be proven wrong. can a Chinese AI researcher go to a conference in the US or the UK exchange ideas with international peers without any scrutiny or hassle from either their institution or from the government? I think that, even if that's possible in some situations, I think this this general chill is not constructive and we need to find a way to transcend that because it's it's not just in China's interest, but when we look at the things that impact all of us, we want elite talent to work on things like air safety and alignment. And my understanding is that lags in China these days.

It's not nonexistent that research. But I think China would really benefit from having those conversations internationally, and certainly the world would benefit from making sure that we share the best possible knowledge on these issues. Fascinating. Thank you so much. Elsa, I'm going to turn to you. We're going to be talking about military later, but I want your insights on whether there are sort of notable government sanctioned initiatives on talent training or labor initiatives . What does your research tell us. So we've seen a strong emphasis on both education in AI and AI education coming from the Chinese government quite consistently, as both Michael and Caiwei have spoken to, there's definitely been an emphasis both from the government and from many leading enterprises on upskilling workers, where appropriate, on identifying talent they can leverage in initiatives, whether that's lower end jobs or more higher end talent.

And so that the Chinese Ministry of Education has also been relatively forward leaning in encouraging and incentivizing schools at multiple levels in China to lead lean in on deploying curriculum for studying AI as early as high school, even in K-12 education, in some cases focusing on early, early education and digital skill sets and literacy, and increasing the study of AI and existing science and math curriculum. So I think that's roughly going to be an important trend to watch and how much this future workforce starts to be cultivated through the combination of local initiatives, public private partnerships and some of these central policies that are really encouraging educational institutions at every level to lean in on education in AI, to build the kind of future workforce and start to close the talent gap, or even enable China to be really a leader in talent going forward. And certainly there are disparities in China's education system as there are in the US education system.

But Chinese policymakers seem to be quite forward leaning and really trying to look at the best ways to develop, develop and implement these these curriculum and to have schools have access to and incentives to really pursue that kind of education. Also leveraging AI tools for education generally. And this is where I think chatGPT like tools and large language models will be will be definitely an interesting dynamic to watch there. When we've seen AI enabled tutoring or enabled customized instruction really becoming a popular phenomenon and a focus on how to have education be more optimized, individualized, data driven approaches to improving learning. And I think I hope that's something where we'll see more progress in in a US context as well. A lot of different companies across both across both ecosystems that are really exploring the potential there of how it can be used as a tool for talent cultivation and training for different different aspects of the workforce.

And I think definitely curious to see where that will go. And at the same time, I think there's the question of just general reforms and how the Chinese, the Chinese government and Chinese enterprises are responding to the reality that I think I've seen the estimates is as much as over 50% of jobs in China could be either replaced or in some sense disrupted by AI. The data and analysis there may vary, and certainly there will be a lot more cases where air is integrated into. Into the workflow or to the job description.

In some respects, as Caiwei mentioned, with creatives who are already leaning in and trying to use AI to enhance their existing existing lines of effort as opposed to being replaced themselves. So I think definitely the question of how to build an air ready workforce and how that's happening through policies, commercial programing or individual initiative and creativity in some cases I think is a challenge that will be playing out in every context, in every country and why should anyone be listening to us when they could be listening to AI enabled versions of us that might be more interesting or witty potentially in the time to come? So I think even in creative and knowledge industries, just the question of how to have AI enhance day to day workflows is definitely gonna be critical going forward. And I think the implications for the labor market and concerns from the government about unemployment and social stability will probably be drivers of very proactive policies to try to encourage that kind of adaptation. Fascinating. Thank you so much, Elsa.

We have an audience question on innational standards and norms. I'm going to turn this question to Michael. As the integration of AI into industries also necessitates the creation of new standards, especially in fields like autonomous vehicles, where tech safety and manufacturing guidelines are crucial. Does China have the capability to influence in shape those international standards and norms? Well, my short answer would be absolutely. And don't take my word for it. Go and look at the international standards bodies for tech over the last ten, 15 years and you'll see Chinese firms on the advisory boards of of all of those bodies.

I think of a company like Huawei that has done this very well. And I think they're very sophisticated at this and have deep relationships with the standards, bodies that are relevant to their business, and certainly they're their peers. Other Chinese tech companies are very good at this as well. This is really important because, for example, in many countries, if there is an ISO standard, international organization for standards If there is an ISO standard on a particular issue, many countries simply take that as law. And so ISO standards become laws by default. And I think China was ahead of the U.S. The U.S. was really neglecting that reality, certainly during the Trump administration that, you have to work through international institutions.

I think the U.S. is now wise to that playbook and is going to be a lot more proactive in influencing global standards. And if you look at the U.S. coalition, countries like the UK, Japan, Australia, India, South Korea, and also, the European Union, it's a big coalition that is going to be very formidable in terms of influencing standards. And if you look at the EU AI Act in particular, much of it is conceived in direct opposition to the way that China has deployed AI. So it's difficult to see those countries adopting Chinese AI standards. Another limitation would just be language. almost 60% of content online is in English compared with less than 2% of content that's in Chinese. So we used to talk a lot about China's inherent data advantage. That doesn't seem like an advantage anymore when you consider, , a model like GPT for its superiority seems to stem from scale and brute compute force on all of the information on the Internet.

But I still think, those are caveats. But to your to your question about standards, yes, I think China is going to be a major player. They know how to do this. And I think there should be international standards that are forged separate from strategic competition on things like air safety alignment problems similar to, this political paradigm that has been reached in the past for issues like nuclear nonproliferation or biosafety. And if we can't get our governments to start talking about that, then civil society and industry should lead and not not wait and start having those conversations, because it's really important that we do have those international standards and norms. Fascinating. Thank you so much, Michael.

Speaking of Huawei, Huawei's recent progress in 5G smartphone. Nigel, there's talk about this interaction between the U.S. restrictions, sanctions and the pace of China's high tech development. Do you think U.S. sanctions actually had any real impact on a piece of China's aid development, or is it too soon to tell? What's your take on that? I mean, obviously the denial to China of Nvidia's A108, A100 GPUs with a data transmission rate of 600 gigabits per second is a potentially constraining factor. But the evidence, such as it is, suggests to date, Chinese companies have been pretty adept at sidestepping these restrictions. And it turns out there is a major loophole in the Department of Commerce, BIS regulation, because it's still perfectly legal for Chinese companies to rent access to a 100 GPOs through companies like ours on the cloud.

So at the moment, it doesn't feel as if, of course, NVIDIA promptly produced chips that were compliant with the new constraints and is selling goes to China. So at the moment it doesn't feel as if in that area it's made that much of a difference. I would say that in many ways a greater constraint for China is the party's desire that I should reflection conform to the party's concept of norms and values. I mean, they've resile from their somewhat in the face of evidence that you really do have to step aside and let the technology go where it may. But I think the the aim is still there. And if you compare it with Ernie, which is probably the best known current large language model the Chinese have produced.

Ernie is pretty restricted in what it can tell you. Fascinating. Thank you so much, Nigel. So, of course, one of the key purposes of U.S. restrictions and sanctions on Chinese technology is its potential use for security and military purposes. Elsa I wanted to highlight your research a little bit, which is fascinating.

Yourresearch really provide this comprehensive look at the Chinese military's recent advancement, its approach to arms control, potential future trends, as well as strategic implications. Can you just share some of the key insights from your study with our audience? Sure. So just as we've seen this emphasis on a revolution and transformation in the Chinese economy, so too is the Chinese military looking at emerging technologies like artificial intelligence for their disruptive potential in defense and security and the PLA is seeing this as a revolution in military affairs, and just as with many aspects of AI things in the near term, maybe incrementally enhancing existing existing capabilities when it comes to crewed missions and near-term applications. But in the long term, there could be potential for more of that transformation that is that is anticipated and seen. If we look back even at the pillars of reaction to AlphaGo initially, and the idea that an AI agent could best human players in a game of strategy inspired a lot of Chinese defense strategists to think what would and a AlphaGo for War look like? And how could I be applied in decision support and to in to enhance commanders decision making in a complex military environment? And of course, between a game and the actual battlefield, there is quite a great distance and we've seen a lot of advances in the use of AI agents in war gaming and including initiatives under the auspices of the Academy of the Academy of Military Sciences or the Chinese Academy of Sciences that look at sort of testing AI systems in these virtual environments for decision support. We've also seen a lot of Chinese companies coming into the fray and aligned with China's national strategy of military style diffusion, working to market capabilities to a military audience, including folks focused on those kind of decision support and battle management applications.

And of course, just as there are in the Chinese economy, in the Chinese military, defense industry and security sectors generally, there there's a wide range of applications of AI and even even where the player may not necessarily be at the forefront of the latest cutting edge advances, the capacity for implementation and applications at scale or for more flexibility and moving faster to deploy some of these systems could be a potential source of advantage. As Michael mentioned, initially, a diffusion is certainly a strength of the Chinese system in some respects, and I'd argue that could be the case in a military context as well, where not just having advanced autonomous weapons systems or being able to leverage AI in logistics or air defense, but being able to really implement this and field, it is critical here. And I think there's definitely some critical synergies between defense and commercial applications and a lot of a lot of enterprises poised to really operate in that nexus. So, for instance, we've seen a lot of emphasis, including during the pandemic on drones for delivery and e-commerce. So too that can be applied in logistics companies like DJI that are known for world leading drone technology as a lot of folks would still rather use a DJI drone than any of their competitors when it comes to the user friendliness day to day.

So to the PLA has looked at using some of these small scale commercial drones, whether in training for urban warfare or logistics delivery to far flung military units and some of the process optimization and intelligent manufacturing that can be applied in a defense industry as well to increase the efficiency and the capacity force for scaling up or improvement in that context as well. And I think in the near term, some of the most impactful applications could be not just those that are visible. So we've seen the pilot deploy drones or unmanned or uncrewed technically system as across multiple domains of warfare. Every single service of the play has experimented with these uncrewed or increasingly autonomous systems that do levels of attack. We are, of course, difficult to verify when you're just looking at a weapons system without knowing the underlying technical specifications. But beyond those very visible interest interests in and swarming in autonomy and even autonomy in hypersonic weapons systems, there are a lot of more less visible but more potentially near-term impactful applications we might see play out day to day, whether that's cyber security and cyber operations or techniques for cognitive electronic warfare or increasingly psychological warfare and influence operations when it comes to advances in targeting more sophistication in propaganda for disinformation and in support of some of these psychological operations objectives, which will, of course, be primarily targeting the US and Taiwan, given the pillars consistent emphasis there.

I think as we look at how the PRC as a whole and the CCP party state and people in particular are responding to Taiwan elections, the use of deepfakes, synthetic media and more tech enabled techniques for information operations will be important to watch there as the leading indicators of just how quickly the public is able to iterate, experiment and deploy some some of these advances. And I think, chatGPT was very much a wakeup call for the PLA, just as it was for much of Chinese industry that we saw back in 2017, the PRC declaring its interest in leading the world in AI and the pilot to aspire to become a leader in its own right. But when an advance like chatGPT that comes out, that has dual use or dual purpose applications and there aren't, there aren't initially, despite Ernie and some other bots and tools coming online, there aren't quite comparable Chinese capabilities yet, though certainly there's a race to catch up now and LLMs, that is arguably something of a Sputnik moment, for lack of a better cliche for the play too. And there's definitely been a lot of just as there was with AlphaGo, a lot of interest, a lot of study, a lot of consideration of how LLMs could be deployed and employed, whether that's in intelligence analysis of some of these psychological operations or information operations applications, as I mentioned, or to enhance decision making generally. So lots to discuss on this front, but I'll leave it there for now indefinitely. So all of these trends we've been discussing in an economic context, including some of the challenges around education, workforce having AI-ready talent, those are also in play in the play and broader military and defense industry as well.

Thank you so much, Michael. I'm going to turn to you next. Are there ways in which US policymakers can minimize or mitigate against the potential risks that Elsa mentioned, how military technologies by Chinese or Chinese parties can introduce to peace and stability, especially in a context of US-China competition or China's relationship with Taiwan? Definitely. I think the US government, the Department of Defense takes this very seriously, and I would recommend taking a look at a book by Paul Scharre at CNAS called Army of None, talking about these anecdotes throughout history, how autonomy is not really a new concept in the military. And so the principles of deployment and engagement don't need to be invented from scratch. And I think the core point, if you look at it, in any public statement that DOD officials make on autonomy is ethics and law, is "don't just go away when an algorithm is involved." And there's a view that we are a long ways away from fully autonomous weapons, not just from a technical perspective, but from a legal and ethical perspective.

There will always be a human in the loop. I think the US has tried to have these conversations with China, and China has not engaged so far. But again, as I mentioned before, there are some aspects of this competition that we should be able to segment out and work together and, embrace in the spirit of of common international law. And I hope that that happens sooner rather than later. Fascinating. Thank you so much, Michael.

And in addition to Paul's book, Our own Nigel also has a fascinating book that makes it the compelling case that China is rapidly becoming a pure competitor to the US in advanced technologies and China's national effort could potentially outpace the U.S. unless significant measures are taken. I highly recommend our audience to take a look in which he also measures concrete steps which Western decision makers can can consider to maintain its competitive edge. We've touched on so many pivotal aspects of AI today, from industrial applications to the role of AI powered military applications, from domestic economic considerations to international ramifications, international standards. It has been such enlightening discussion, and we hope our conversation, our discussion has given the unique new insights into how AI is reshaping China's economy and its military. Big thank you to our panelists for sharing their wealth of knowledge and to our audience for your participation.

Please stay tuned for future announcements on Asia Society Policy Institute, Center for China Analysis's website. This is not the final installment of this series, we will be inviting more panelists for our future discussion. Thank you so much for participating.

2023-10-07

Show video