Kat Duffy (3:17)

Hi, everyone. Good afternoon. If we could get seated, we'll hop into the last panel of the day. I send out extensive notes before panels that I moderate. And I would like to take this moment to thank Sam and Selena for following our dress code of black and white. And Jeff and Mika, I don't know what happened. You know what I mean? It was really right up at top, just black and white in honor of the AI China, US Discourse. So with that, everyone, welcome. I hope everyone has been having a wonderful day at this wonderful forum. I'm Kat Duffy, senior Fellow at CFR for Digital and Cyberspace Policy, and I have the pleasure of moderating today's panel on tech rivalry and competing visions of the future. We are joined by four exceptional panelists, none of whom I actually gave a dress code to. And I'm going to ask each of them to quickly introduce themselves because there is nothing worse than someone reading you a bio you can already read. And so with that, Sam, can we start with you?

Sam Sachs (4:18)

Thank you, Kat. Sam Sachs. I'm a senior fellow at New America and Yale Law School. I've worked on Chinese technology policy for the better part of 15 years, both in the national security community as a linguist and in the private sector. I'm writing a book now on data and trust in the context of US China relations with University of Chicago. Next year, hopefully, we'll be out.

Jeff Ding (4:42)

Hey, everyone. My name is Jeff Ding. I'm an assistant professor of political science at George Washington University, just a short bike ride away.

Mika Ouyang (4:50)

Hi, I'm Mika Ouyang. I'm a visiting professor at Carnegie Mellon University and formerly of the department formerly known as Defense, where I was the Deputy Assistant Secretary for Cyber Policy before that, with a long career in both think tanks and on Capitol Hill working for various national security committees.

Selena (5:08)

Hi, everyone. I'm Selena. I lead China and AI policy work for the office of Eric Schmidt. So great to be here.

Kat Duffy (5:15)

Fantastic. And thank you all for being pithy. I appreciate it. So I want to start with a first popcorn question to each of you. I want really one or two sentences, Max. We've heard today on panels that discuss what China is looking for, what the United States is looking for. The word rivalry features very heavily in the discussion of this panel. Is that the right frame? Do you think that rivalry is part of where we should be centering this discourse. One or two sentences, and we'll start actually with Selena and come backwards.

Selena (5:49)

Yeah. I think the word rivalry kind of connotes a lot of different assumptions. Right. One is that it's more of a zero sum kind of competition, that if one side gains advantage, the other side kind of loses out. In AI, it's not exactly clear, I mean, various technologies, but I think AI is the one that I focus the most on. It's unclear if there are destabilizing impacts from AI that happen in China that the US would be immune to that. So I think one, that's one of the problems I have with the term rivalry or to the larger extent when you talk about an AI arms race or race to AGI. The other one, I think is whether people in both the US and in China have a desired clear end state that they're both trying to get to is also unclear. We don't know know if China is also trying, for example, to race towards artificial general intelligence, if they really believe that there is this permanent decisive advantage that can accrue once you reach that state. And then I think the last one that I would be curious to hear other people's thoughts on here is whether the US and China are doing similar approaches to AI. If both of them are really trying to embrace, for instance, what we see in the United States and in Silicon Valley, like trying to scale their compute in order to get to ever larger and more intelligent models. And if that's really what China's strategy is, whether by necessity it cannot, for example, because of compute, or whether there is another strategy that they're approaching. So unclear if they're running in the same kind of lane at all and trying to get to the same end state.

Kat Duffy (7:24)

Mika?

Mika Ouyang (7:24)

Yeah, I think rivalry is the right term just because whether or not it should be the right term, whether or not that accurately describes the relationship, there is certainly an amount of looking at the other side's goals, not just on AI, but the broader technology landscape that they are trying to think about surpassing each other in different ways. That one side defines the goals in a certain way, the other side says, I can do that too. And they look and sort of pay attention to each other in the ways that I think siblings with tension have, that there's a sense of it's not just one side is ignoring the other. Right. It very much is a rivalry, even as there's a lot of interdependence underneath that.

Jeff Ding (8:10)

Yeah, I think I agree with that framing a lot. I think to Me, in a lot of my research, I think it's not that competitive in that I think the US Is very far ahead when it comes to being positioned to win the AI competition. But where the rivalrous element comes in is as long as both sides believe that they're trying to benchmark against each other and that they're in this strategic competition, then you still have a rivalry. Like I went to University of Iowa for undergrad. I don't think we have a rivalry with Nebraska, but our fans believe that we have a rivalry, even though we've been beating them every year. But as long as there's that perception and belief, I think the rivalry continues.

Sam Sachs (8:59)

I am going to answer you in one sentence. Rivalry serves specific actors and specific interests.

Kat Duffy (9:07)

I love it.

Sam Sachs (9:08)

All right.

Kat Duffy (9:08)

And with that, we're going to go to some what some of those specific interests might be. Mika, you served as. I think this is a fair framing. The senior civilian responsible for cyber policy at the Department of Defense. You know the security side of this as well as anyone. But you've also argued that the security conversation is crowding out some elements of a conversation around economics, around innovation. And the FCC just issued a ruling that would effectively bar most foreign made routers from the US market, which could include routers like from Cisco. And so walk us through what you think that type of ruling would actually cost, what the security case actually is, and where high walls make sense versus just disservice.

Mika Ouyang (9:56)

Yeah, I think that one of the things that we worry about and coming out of the cyber world you worry about a lot, is that the devices that empower communication, the Internet and technology development throughout the world are somehow compromised or compromisable in the interest of one state or another. And it's this suspicion and fear that drives people to say we shouldn't have. We can't trust products or software applications that originate from one or the other or even made there. But that suspicion and fear doesn't take into account the reality of supply chains and how things are actually made in a technologically advanced society. That when you look at the ways subcomponents are made, where final assembly happens, when you look at really the way the economic benefit is put together for, for semiconductors and everything that gets built upon that, it's a much more complex picture of how that all comes together. And so when you start saying, hey, we can't have Product X from a particular country or we can only trust things that are made inside of the United States, you're really worried about one type of compromise which is that it's been engineered in to the advantage of the person who's doing the engineering. You're not worried about, hey, this product may just be crappy and someone could take advantage of finding a vulnerability in it later to compromise it. But in banning product as the FCC has done, and actually they're saying it's only a one year ban, you're sort of springing this on companies that have had their supply chains developed for a really long time and saying like for the next year you can't sell this product into the United States if a certain percentage of it has been made internationally. For everybody who's like trying to set up Internet at their house, like the home router that you get, like, you're not paying attention to what percentage of that is made inside the United States. And if you're a company that's prepared to ship orders to people, like, the idea of a year long suspension of your economic activity could result in very serious economic harm to a whole range of companies without really understanding what percentage of routers actually would meet the threshold to be unaffected by that and then what the size of the market is. So there's this lack of economic analysis that's coming when the government starts taking these steps to try and address the suspicion and a lack of understanding of what that means for all of the innovation that is empowered by those devices and those things. And I think it's quite destabilizing to the entire technology industry.

Kat Duffy (12:41)

So Jeff, I want to move to, to you now because what Mika is thinking about is a global ecosystem, is a global business chain, global markets, and you write a newsletter that has 30,000 subscribers at this point each week. And so anyone who's not subscribing to Jeff's newsletter, I would send you his way. You've argued that competition can be understood as a diffusion marathon and so not just a technology sprint, not just sort of an AGI sprint. Can you talk through how you see the actual race that may be occurring and how would that change the way that we're approaching things?

Jeff Ding (13:30)

Yeah, I think what's beautiful about rivalries in sports is you know, who's winning. Right. There's a clear endpoint. What gets complicated with discussing this US China tech rivalry is we often have to be very clear about what we're racing over. And then how do you measure that? And so, you know, we have so many panels, events where people are just saying things and it's like, oh, competitiveness or China's, this industry is better. And for me, when I look at the rise and fall of great powers. Ultimately what countries compete over is productivity growth. If you look at all These patterns of First Industrial Revolution, Second Industrial Revolution, U.S. japan, competition in the information revolution, the country that capitalizes the best on these technological breakthroughs is the one that's able to diffuse a general purpose technology throughout its entire economy and achieve productivity leadership to sustain economic growth in the long term. That sort of economic power then translates into other dimensions of the rivalry, such as geopolitical and military influence. And so for me, I don't think that should be the only thing that characterizes the U.S. china rivalry. But I think whenever you're trying to make some claim about US China competition, give me what the end outcome we're competing over is. Be very clear about why that matters. If productivity growth, productivity leadership, economic leadership is your end outcome, then it becomes clear that the competition over AI is about which country can adopt and spread and diffuse AI throughout its entire economy, not just its frontier firms, but it's small and medium sized businesses in Iowa or a county town in Qinghai Province. And to me that's the crux of the US China AI competition.

Kat Duffy (15:36)

I'd like to take moderator's privilege here to emphasize that between the University of Louisville and the University of Kentucky, no matter who is winning, the University of Louisville is winning. Go Cards. Selena. I want to feed off of what Jeff had to say to go to you. So you have said many times, are their interests really that different? And if you were to take Jeff's lens for example, as well, that really the end goal of this, the reason that the AI transformation matters is because where it leads is in economic stability, in global power, in global dominance. Do you agree that that would be a more constructive framing? And if so, how do you see that playing out in both nations?

Selena (16:20)

Yeah, I agree with Jeff. I think there has been an extremely heavy emphasis here, especially coming out from Silicon Valley and the last admin, about trying to reach AGI, which is imminent essentially for people who don't follow this artificial general intelligence would match human level intelligence and that would unlock huge gains and this would bring abundance, prosperity, huge takeoff moment. And once that happens, this gap will keep exacerbating and this would not allow any other country to catch up. So I think that that has been the thinking that has been very pervasive. But if we adopt like what Jeff just said about like a diffusion marathon, which I think I greatly agree with what China has been doing because it's maybe by necessity, because it hasn't been able to access as many AI chips as the US has been because of export controls, they've been very much trying to lean into their own natural advantage, which is to integrate AI into manufacturing, into industrial transformation, into existing areas where they already can try to accelerate so hardware. If you think about what they're trying to export, you can think of robotics. So that's another emerging area. So I think that has been what China is doing and if you look at what the US side is doing, it's a lot more enterprise adoption software. So both, both are kind of doing, maybe trying to lean into their natural advantage here and it's kind of complementary. But sadly because of frameworks like rivalry, because of the sense that if one side gains, the other side loses, it kind of precludes both countries from doing any kind of collaboration and cooperation that can even seem constructive and like quote the Chinese term win win from the panel earlier. So I think there is, it is necessary to shift that framework a bit and to think of maybe what is national competitiveness if it's not just an innovation moment or an AGI moment, if it's more of a diffusion framework, then there are areas where it can be off limits and close to national security, like military advantage and application of AI. But there are also areas where it would really benefit from collaboration and both sides coordinating on certain areas.

Kat Duffy (18:47)

So I mean I think what I hear from both of you is essentially it is completely fair to say there are competing visions of the future, but it's not inherent to have in competing visions of the future that those box out any space for collaboration or shared visions for, for some elements of the future as well, that there might be space for both. Sam, I want to turn to you. You've been working on Chinese tech policy for over a decade. You're fluent in Mandarin, you're writing increasingly on AI. When you're really looking at this from the Chinese lens, from what you're seeing on the ground there, from their own incentives and drivers, what is resonating for you? Where do you think we're over indexed, especially in D.C. on the discourse and where do you think we're maybe under indexing?

Sam Sachs (19:33)

Leslie on the previous panel said the United States has many actors and many interests and it's important to disaggregate those. And I think it's also important to surface that so does China. And so when we say what are China's intentions? We have to ask, we have to look at many different Chinas, even in a top down authoritarian system. And I want to give you an example of this, I was in Shanghai last summer for the World AI Conference the same week as America's AI Action Plan dropped. And one of my counterparts in a dialogue came in and she had circled in America's AI Action plan all of the references to the Pentagon and to the intelligence community. And she said, see, the US is now full on into militarized, securitized AI. This makes it very hard for those of us in China's system who are pushing for more guardrails to have space to advocate that. And I worry that we create a self fulfilling prophecy by flattening China into a monolithic other. Yes, I am sure there are leaders in China's system that very much are part of an AI race mentality. But there are also a lot of other interests and stakeholders. And that's just one. If you look at China's matrix of laws and regulations that touch on different parts of AI, you will see what my dear friend Kendra Shaffer at Trivium calls the sweet and sour of Chinese law. You'll look at China's measures on generative AI and right up there, anything will be prohibited that undermines state stability and the ruling power of the party. Okay, got it. You read further down and you see things that actually mirror some of the concerns that many civil society groups in the US are talking about. Actually today, China's Ministry of Industry and Information Technology released new measures focused on ethical reviews, where companies will have to file an ethics review along with their algorithms. What's one of the provisions there? Well, they talk about preventing algorithmic exploitation of workers so that humans can come in and sort of take control when an algorithm is veering in a wrong direction. These are just a few examples, but I highlight it to say there are many different groups and debates happening in China and I think it behooves us to pay attention to those debates. One more example and then I will get off my soapbox on this. You know, China released the first smartphone with AI embedded in the operating system in December. It kicked off a fierce debate about who the expansive permissions that come with agentic AI, who owns data, who controls it. Many of those concerns have been mirrored in the us. The CEO, the president of the Signal Foundation, Meredith Whitaker, was on the stage at By Southwest a year ago sounding the alarm about these very same concerns about the expansive permissions that will come with agentic AI. So I think we have to pay attention to those debates and not always zero in on the great power competition lens if we want to move in a more constructive Direction.

Kat Duffy (22:50)

That is a great segue into my next question, which I'd really love to hear from each of you then. Where are the spaces? What is the biggest loss when we frame this as a sort of a hard rivalry and we securitize the conversation? So there's climate modeling. It could be AI for science. It could be materials sciences developments. It could be in pandemic response, pharmaceutical innovation. There's all sorts of areas where the world's two leading AI production powers could be working together to fundamentally transform things. By the same token, like, I live here on Capitol Hill, and you're looking at cybersecurity attacks, vast attacks, not only into our critical infrastructure, but into companies like Claude, into our leading frontier companies. People will talk about corporate espionage. People will talk about authoritarian concepts being built into the AI models. People will talk about model diffusion. So essentially, Chinese labs, basically, in an unauthorized fashion, building what they build off of billions of dollars of American investment. Where do we want to really keep a very low fence and drop barriers to collaboration? And where is a high fence merited? Where is there a true rivalry or a true security concern that we should have in mind and that the United States would be justified in holding? And that's to any of you, and I'll let any of you start with it.

Mika Ouyang (24:33)

Yeah, I mean, I think in the Biden administration, we talked a lot about this, like, very small yard, very high fence, and we tried to define the yard in a really small way. But as we look at the development of export controls in the United States, which what happens is that as you move away from the specific end product, which is sort of where we started this conversation in the 90s, right, we were talking about supercomputers, we were talking about, like, actual boxes that would be shipped around the world. And we move further and further into the stack. We started talking about the semiconductors, we started talking about component parts to build the semiconductors. We started trying to prevent the diffusion of technology in the ability to make the technology. And then we started talking about talent. And so, like, as the suspicion continues to drive, we start securitizing more and more of this conversation. So the idea of, like, hey, if we want to try and prevent them from getting the highest end, if the US Wants to try and prevent China from getting the highest end semiconductor chips, well, the way that we have to do that is prevent them from having access to the any part of the supply chain. And so the yard gets really big really quickly, and it's very difficult to try and prevent that from happening. And you look back in the history of this in the 90s, in the Clinton administration, and even with all the export controls you had on supercomputers, you were still seeing some of them wind up in places in China, in Russia, where you didn't want them to be. But as technology advanced, and it advanced very quickly, the technology that you were trying to control one month was the technology that was available on the corner store the next. And so like, they were not able to see the restriction, you know, to put restrictions into place in any kind of useful way to guard against the ways in which technology itself was developing. Like, they would talk about like, well, a computer that would allow you to do weapons development is at what they would then call like 7,000 MTOPS, millions of theoretical operations per second. My MacBook Air is at 14 trillion theoretical operations per second. And that's like a commercial product. So you can't. When you think about how you're going to make that fence and make that fence high, you're also trying to make this fence against a rising tide of technology development, which makes it almost impossible to do.

Jeff Ding (26:57)

Maybe I'll say one thing on this over indexing versus under indexing thing. I do a lot of research on great power competition. It gets clicks, it gets readers. I think we should have really good research. We should have a lot of time and resources going into how to more effectively compete and what the true nature of the rivalry is. I just think right now, if you're looking at it in terms of a marketplace of ideas, there's plenty of supply for that. I could go to any think tank in D.C. and they could give me a good rapport on how do we beat or compete with China on electric vehicles or something, or batteries. There's stuff I can find that, but what I can't find is actually examples of what Sam just gave us. Wow. There might actually be interesting convergence between the two countries approaches to AI governance. There's not as much market demand for that. You do a report, this is how you beat China at something. It can get cited by Congress. It gets fun. You know, there's structural incentives that support this sort of framing. And it's important to have good research there. But just if you're a student in the audience, if you're an up and coming researcher in the audience, one way to distinguish yourself is to sort of do something that's relatively neglected and sort of research things that are relatively neglected in this field.

Kat Duffy (28:32)

Same or Selena, any thoughts?

Selena (28:33)

Yeah. I also want to say, so, Helen Jiang from our office is here and I'm sure our office would love to fund projects related to so just putting it out there. If people want to work on this, speak to Helen after. But I pour one out for Helen. So I do think like I, I think it's, it's hard to think imaginatively right now about like oh, collaboration but just talking about the reality here of, of like how intertwined the two AI ecosystems are even despite all the, all that has happened right like the over securitization. Like a lot of American startups are built on Chinese AI model like open source models. Like I think there was some A16Z statistic about of like 30% of startups are built on open source and 70% of those who are built on open source are built on Chinese open source models. And then Europe's last year which happened in the US I think over half of the researchers who presented their studied did their undergrad work in China. So I think there is a lot of existing academic collaboration. AI is extremely porous. Like it's so hard to really fence it off that if we try to get on the slippery slope of saying hey, we got to constantly keep this gap, we have to constantly ban this new technology as it advances. The way Mika was talking about it, I think we are just stuck in this doom loop and it will keep spiraling. But once we stop and say hey, okay, maybe we want to stop being in this cycle and then it gives us a bit of breathing space to think of hey, what are some things that maybe people could collaborate on? And there are some very low hanging fruit. One would be, I think what Sam just talked about, like when she talked about the. I think you were talking about ZTE by dance phone. That's like agentic. Yeah. So like agents like openclaw, which is an AI agent has been absolutely like phenomenally popular in China. Like people talk about raising lobsters. Like people lined up outside Tencent's headquarters trying to get employees to help them install it on their laptops and they buy things on Taobao like E commerce and try to get it to install even if they don't know tech. So there is this real fervor and hunger for good AI technology regardless of where it's from. But the problems are if you just leave these AI agents which are autonomous, just running amok without interoperable standards between the US and China, there's going to be a really dangerous situation. It's happening in the next six to 12 months. So there are some things that I think it would really Help both ecosystems if they come together and just talk about coordination. It doesn't even need to be shared standards. It can be like, hey, our standards are parallel and interoperable. That's one. I think the other thing would be just like preventing malicious use of AI by non state actors. Right. I think both countries have a shared interest in that. And if we can even try to do these coordination efforts, then maybe we can try to progress on some actual collaboration things and like AI for science and climate change and maybe like measurement science, which was in the White House, AI action plan evaluations and thinking about interpretability of models. So there are a lot of areas, but there's just such little trust and we're still kind of stuck in this old doom loop of like, oh my God, we need to slow China down, that it doesn't even allow for anything that can be remotely productive or constructive.

Kat Duffy (31:56)

Sam, how about you?

Sam Sachs (31:58)

One of the things I've been thinking about is how do we find these space for collaboration amid low trust? Because while some of us in this room may have a level of trust based on their own experiences, I don't think that's going away. Can we think about, for example, what are the technical and what are the policy or legal solutions that can enable connection without trying to build trust? So what do I mean by that? And these are going to be very. I've been trying to think about what are very concrete, even narrow areas. You know, are there ways to share kinds of data sets that are anonymized, encrypted in very narrow areas related to AI incidents that wouldn't compromise sensitive or proprietary information? For example, in the area you mentioned, environment. Right. Are there ways that, you know, carbon emissions data or even understanding patterns of EV charging stations? I know these are very specific, but I think that's the only way if we're going to think about concrete areas. Given the fact that I do not think the trust issue is ever going to go away, but it's going to require stripping out some of the politicized areas. The one other thing I want to highlight is what is at stake if we don't do this right? And this is also where the conversation gets bigger than just US and China. This is a moment where it's an open secret in Washington that if you want to get anything done domestically, you just tie the word China to it. And particularly at a moment where there's a fierce debate about whether and how there should even be guardrails on AI, I think that there's a lot at stake if we go down a securitized path where maybe we're not necessarily making the data or the proprietary technology any safer or more secure, but we are entering an environment where this comes with real vulnerabilities and it's so much bigger than any one country.

Mika Ouyang (33:58)

Yeah, Kat, before we leave this topic, I think that there is a real opportunity here. You asked about, like, where are the areas for collaboration? And especially because Selena made the point about the diffusion of Chinese models into the global ecosystem, there is a place where the US and China could be working together very productively as they develop even competing visions of AI. And that's on cybersecurity. Because as the Chinese models, which are open source, continually diffuse through the world, as the US Frontier models get out there, there is this place where Chinese models will run on American devices, American models may run on Chinese devices. Being able to have a global conversation about where vulnerability is found and then how to address it is really important. But right now, China has a policy that requires disclosure to the government first, as opposed to the US System of vulnerability disclosure, which has a whole independent researcher community where there are bug bounties, where people find things, where they report it to this common vulnerability database so that people can find it. And then you have this whole industry that's around figuring out how to patch it and address it or look for these things. But that's not an ecosystem that is shared in China. And so without that ecosystem in China, their products will continuously be less secure because they don't have this whole crowdsourced security environment, not in a militarized sense, but in like actually preventing malicious actors from taking advantage of the technology. And if the two can't work together on this, and with the rest of the world, as the rest of the world is trying to decide which set of products, maybe take some of column A and some of column B, we're not creating a safer world where the technology is trusted. It.

Kat Duffy (35:45)

So I'm going to push on this a little bit because by this it's. To Sam's point around low trust, I would say cybersecurity and shared value for cybersecurity and what is and is not acceptable is probably not a highlight of the relationship right now. But I also think it is

Sam Sachs (36:08)

the

Kat Duffy (36:08)

narrative of look, what they're doing to us, which is always absent. The narrative of maybe we're doing the same thing is tricky. And there's a long and healthy history within great powers of quid pro quo on some levels, and then on other levels saying, okay, we are going to agree that we don't want this to get too far out of hand. We're going to do this for specific purposes. We each kind of know the other is doing it, but we can also have a shared space where we diffuse a better, healthier set of norms generally. So, Mika, I want to return to you on this point. What is your. You have that narrative and then what is your rejoinder to the people who come back and say sulfa tune, like the agent Aki, you know, attacks on Claude Volt Typhoon, Chinese incursions into critical infrastructure EVs, they're here to take us over.

Mika Ouyang (36:58)

Yeah, I think that, right. President Trump was confronted about this several months ago and said like, well, what do you think about Volt Typhoon? And he was like, it's a nasty world out there. You don't think we're doing it to them? And like, it's the first time that you've really seen a US President say, like, hey, you know, we're not just playing the victim here. We're giving as good as we're getting. And so I think that there is this sort of, like, presumption amongst a lot of cybersecurity professionals of like, oh, the Chinese are getting us everywhere. And some of the things that are happening is, a, the US is really, really good at what it does, but B, China does not have the same attitude towards transparency that the US does in cybersecurity vulnerability. So we have this whole cyber industry that's hands off or at arm's length with the government that is sort of finding and patching vulnerabilities that maybe would be of advantage to a national security person to leave that back door unlocked. But we have this sort of idea that you take the technology world as you find it and we try and prioritize security in that conversation. We don't have that same, we don't see that same conversation happening in China. And so there has to be this understanding that the two countries have certainly the capability to do to each other mutually what is being done to them. But also that the US name and shame approach might be one that it does not like when turned around to them. And we saw a few months ago, for the first time, China release a technical attribution report on what it claimed was an intrusion by the National Security Agency. The US Government, of course, has not acknowledged or confirmed any, you know, not said anything about this. And it's entirely possible that this is something that, like, there are tools and techniques that were out there. Who knows what the truth of it is, but it is the first time that China has put something out there saying like, not just, no, you're the worst, but like, here are the reasons why we think this was the United States and that could potentially have real consequences for the United States. The United States really suffered big diplomatic impacts when we saw the Snowden Leaks, when we saw WikiLeaks and through Chelsea Manning, that, like the idea that the US Is out there conducting this type of activity really undermines US Trust as well. So there is this sort of reciprocity issue, this mirroring issue in the cyber world that we haven't really developed the kind of norms around that we did in say, the Cold War with like in person espionage. And we sort of had rules about how we would handle those things with an understanding of mutuality.

Kaiser Kuo (39:47)

Let's take a quick break to hear from this week's sponsor. We'll be right back. These days there's no shortage of MBA programs. What's harder to find is one that still feels connected to the real world, to people, to place, and to the way business is actually changing. The MBA at the Hong Kong University of Science and Technology has that connection. The campus sits above the South China Sea, and yes, the view really is as good as they say. But what makes it special is the mix of perspectives it brings together. Students come from everywhere and the faculty don't just teach business models. They push you to question the assumptions behind them. If you want to understand how China's economy works on the ground and how it fits into the larger global picture, HKUST is one of the few programs that can genuinely offer that you'll graduate with more than a credential. You'll have a clearer sense of how to lead in a world that keeps surprising us. Find out more at MBA Hkust. Edu HK so I want to pick

Kat Duffy (40:53)

up on mutuality and this will be my last question to our panel and I want to open it up for questions from the audience. One of the things that I found really interesting about what I'm seeing China do is that they're not only leaning into what they are best at, like manufacturing, but they have also been increasingly leaning into what the US has traditionally been best at. For example, funding university research, building strong trust and narratives of aligned values with nations all over the world leaning into multi stakeholder and multilateral fora as opposed to pulling out of them. And so on the one hand, we have this current narrative that the way that we will make the United States the most dominant force is essentially by centralizing power and possibility in a very few actors and hoping they prevail globally. And on the other hand, we have China actually approaching this topic with a greater focus on research on openness, on global cooperation. And it's someone who spent years representing the United States. It feels a little funky. So I want to ask you all, when we think about a competing vision of the future, where is our competitive strength? What would it mean to out compete China on our terms by Americaing as hard as we could on the things where we would truly be the greatest competitor and create the hardest rivalry for them to beat?

Sam Sachs (42:33)

Sam, I want to pick up on what Mika said about how there's a also a deficit of trust in terms of using US technology. And I wonder if we can sort of flip the prevailing view on its head. Right. There's a sort of view that in order to win the AI race, there has to be a race to the bottom in terms of guardrails. Is there a commercial advantage to trusting the technology that you're using? And I think in part, you know, can we have a race to the bottom on privacy, for example? Maybe some would say this is naive, but it's worth, it's worth looking at because if you go to some of the third country markets where US and Chinese firms are competing toe to toe, I think something that you might hear on the ground is, well, I'm choosing between US and Chinese surveillance. So which is going to be, you know, give me the most bang for my buck. Right. And so how can we think about building in standards and the technology that might make it more appealing at a moment of rising AI sovereignty concerns around the world, where there are increasing questions around the relationships between governments and firms, not just in terms of in Beijing, but in Washington.

Mika Ouyang (43:51)

Yeah. I mean, you see this in the demand for Apple products. Right. On the trust issue. Right. The idea that Apple was resisting the FBI, that Apple was putting end to end encryption on its devices so that nobody could get in other than the user, was the idea of US competition on privacy, on security, on trust. And I think that that's one of those areas where you really do see like that arm's length relationship that the US has traditionally between industry and government really coming in as a commercial advantage that you don't see in Chinese products because of the way the government demands certain kinds of access, because of the way the government has not allowed that trust ecosystem to grow. And we have seen that when you do tear down analyses of Huawei and ZTE products, that they're just not as secure and so there is this issue where the US could be competing on trust. It would require a slightly different orientation where the US Would have to prioritize that. But it is one of those areas where it's been traditionally a strength for United States. We did see for the first time at the end of the Biden administration, the FBI actually call for people to use end to end encryption in response to the Volt typhoon and Salt typhoon attacks. Before that, they had been running around telling people like, build us in our own private back doors. And like, you know, that wasn't going well for them. But, you know, this, this idea that the FBI would say, no, everybody use end to end encryption shows a very different approach to that. But Kat, to your point on, like, what could we be doing to. To like, be our best America? Like, you know, I, I just drove in this morning from Pittsburgh and like, Pittsburgh, Carnegie Mellon is the best AI school in the country. And the amount of federal funding that it has lost as a result of DOGE cuts and other things like, these are funding researchers who are going to be doing things at the, at the cutting edge of what we need, not just in agentic AI, but in embodied AI in robotics. We draw the best students from all around the world to come to Carnegie Mellon. But we've seen a tremendous drop off in the foreign students coming to the United States who come because they want to be here to study with the best people to stay. America's the only country on the world where you can come from anywhere and become an American. It's not a race, it's a creed. And like, I am here in this country because there was a time in America where we decided we had to win the space race. And so we took the best minds from all over the world, including my grandfather, who built the retrograde engine on the bottom of the lunar lander and helped work on the Polaris missile. And like, we. And they took actual Nazis and we decided that we were gonna win a talent race. And like, that. That's what America can do if we're actually serious about winning this. It's not just about the fabs. It's not just about the trips, it's about the talent and that investment. And like, that's where, if America was really gonna lean into that, we would take a totally different approach on education, on visas, on research funding, on all of these things. And, and that's where the long term advantage is for the United States.

Kat Duffy (47:15)

Jeff and Selena, I want to give you both a space to say, like, if you could see The US really leaning into one or two major things right now. What would you want those things to be?

Selena (47:27)

Yeah, I mean, so agree with what Mika just said. Like, I think there are some very obvious things, you know, like welcoming talent, funding, like foundational R and D, which has traditionally been like a major US strength and source of dynamism and going a bit into AI. I think it's kind of ironic that US is not leading on open source. I think they need much. Either the Frontier Labs should put out more open source models or we should just support more open source efforts. And I think there are some like for instance, like Nathan Lambert's doing the Atom project. So I think there are some emerging strands here that would be very helpful and just help accelerate adoption. Right, because developers, hobbyists, academics, these are people who rely the most. Small medium enterprises, these are people who rely the most on open source models and would benefit. And if the Trump admin is serious about like we want adoption, then open sourcing really helps. And it also helps to project American soft power. If America is so worried about Chinese open source models being the new digital belt and road, then put out our own open source models. So there's that. But I also think it's a bit about the narrative and the vibes also because there have been a lot of surveys and the Chinese average citizen is so much more optimistic or trusting of AI and enthusiastic about AI improving their lives. Like, I think that was a recent poll from like a month or so ago and it's like 82% of Chinese are believe that AI will make their lives better and that they trust it and that it will solve certain issues. And in the US the number is like 32%. So I think there is a huge gap here. And one reason is because the narrative that's been told to everybody in America is that, hey, AI is going to replace all your jobs, it's going to make your kids like mentally ill, it is going to build data centers in your backyards, it's going to increase your prices. It's a terrible narrative. Why would anyone want to use AI after hearing that? So I think there needs to be a different kind of narrative and it would help with very positive public examples of how AI would actually improve people's lives positively. And you don't need to wait until AGI comes for that to happen. The existing diffusion of existing AI capabilities is already not that great. So even if you just use existing AI capabilities, a lot can happen. And I think more of that instead of saying, oh my God, if anyone Builds it, everyone dies. I think that would really help in the US And I think that would resonate more with governments and people around the world as well, who feel very detached from this Silicon Valley bubble. Right.

Kat Duffy (50:15)

So we would need to change the narrative. We'd also probably need some policies that would help people believe that the narrative wasn't going to be true. Jeff?

Jeff Ding (50:22)

Yeah, I think my fellow panelists have given a lot of good ideas about America maxing. Right. There's all this, the Gen Z people are like China maxing here. This is America maxing for eagle maxing. Yeah, Eagle maxing. You know, to me, I think I want to go back to. I think sometimes we think of cooperation as the exception and sort of like we're in this world of decoupling fragmentation. But look, we can point out these isolated examples of cooperation. And for me, it's like, if you look below the surface, maybe a better way to frame it is cooperation is the norm. This thing like the WI fi ban is the exception. And then there's all these great ideas for cooperation. I think there's just less of a hurdle to get those across the door if we have this better beneath the surface view at it. Let me just give two concrete examples on things we've just been talking about. One is this vulnerability disclosures thing that Mika was saying. Well, one of the most important vulnerability disclosures that's happened in the past five years was this disclosure of the Apache log 4J bug. Guess which organization made that disclosure? Alibaba. That led to some of these restrictions on Chinese companies having to report to the government first. But Trey, her and others at the Atlantic Council have done a report and shown that Chinese contributors continue to be a main source and a growing source of vulnerability disclosures to common global public goods. Okay, second thing, we've just been talking about encrypted communications. Snowden leaks come out. Okay, us dominating these choke points and have leveraged the Internet for their own purposes. We should see fragmentation. We should see no more cooperation between the US and China on this front. What happens? Well, engineers get together at the Internet Engineering Task Force and develop the Transport layer Security Protocol 1.3. If you're not listening to me right now and scrolling on your phone, check your URL. You should see an HTTPs in front of your URL. Maybe double click it to get the full URL. That's a product of the post Snowden leaks, where US representatives, engineers from around the world, including China, come together and come up with a better encrypted protocol to prevent at least provide another check against government mass surveillance. And it's received some uptake in China in some of the numbers I've been tracking. And so to me, that's a starting point. Cooperation is the norm. Cooperation amidst rivalry is the norm. These things that we hear in the news all the time, the things that make the policy headlines, that's the exception. And then we can have a very, very different conversation about US China, competition.

Kat Duffy (53:03)

It's really helpful. Okay, I want to open it up to questions. Oh my gosh, so many questions. Okay, let's you two just hear in the front maybe. Yeah, we'll start here, maybe batch all three quickly, try to do fast responses and then go to the next. Hi, my name is Asia.

Mika Ouyang (53:26)

Quick question. It's a bit of a political question,

Selena (53:29)

but still I think we can't forget this moment. How have China's technological and military capabilities evolved since the 2023 surveillance balloon incident under the Biden administration?

Kat Duffy (53:44)

Okay, we have surveillance balloons. Kristen Cray, A bit of a loaded question.

Mika Ouyang (53:49)

If 20,000 people could meet, young people could meet at the National Convention center next month to discuss AI talent, races or cooperation. What sorts of things should they prepare for? It's a loaded question because Selena's foundation is hosting that under AI Expo, SCSB AI in Central Cheming in central Michigan.

Selena (54:14)

A Chinese company, Goshen, tried to build

Mika Ouyang (54:17)

a battery component manufacturer in central Michigan, but the effort failed partially because the residents there said, oh, we don't want this Chinese company in our backyard. Maybe a mostly white country like Canada would be okay. People don't remember that Michigan was where a man named Vincent Chin was murdered. A Chinese American who many of the auto workers, I guess, thought he was Japanese, but this was a result of the anti Japanese car imports decades ago. I just wonder, nobody said this in the entire conference.

Jeff Ding (55:00)

What role does race, resentment, jealousy play

Selena (55:05)

in this whole debate between the US And China?

Kat Duffy (55:09)

Okay, I will open it up to all three of you. So we have how capabilities have shifted since 2023, where we would want youth in particular to be coming together around a question. And that might in fact be a really great question for a bunch of youth to be discussing. But also to what degree Race, but then also xenophobia. Right. Could be, could be playing into this discourse as well in the general populace. I'm going to ask for quick answers just so we can get to the other question.

Mika Ouyang (55:37)

So I would just say on the, on the technology development since the balloon, I would just sort of say the technology developments that have happened since 2023 in China were probably independent of the balloon. Right? China sets these plans. They, they set them in, in a long, in advance. So I don't know that things would have changed specifically in response to that incident. And there are a number of things that have happened in their technology race. But I kind of want to decouple those two things. But one of the things about the balloon that I think is really significant is just like the lack of crisis communication mechanisms between the US and China to deal with fast breaking things. And it did force Biden and Xi to talk to each other. But like the difficulty of being able to speak to each other quickly in a crisis still remains a pervasive problem between the two.

Sam Sachs (56:24)

I would love for young people from China and the US and around the world to talk about AI and education. As a parent, that's something that I witness every day in my household and I've talked to parents in China who have very similar concerns. And I would love for my 13 year old son to talk to his peers in China about how they're thinking about AI in schools.

Kat Duffy (56:47)

And I'll pull on Jeff and Selena's threats to say on the last question at least that where Jeff was talking about how there's not research that is currently being done or more research would be helpful. I think the question that you asked is probably a very important one to be researching. And I think the political costs of proposing that research have increased dramatically in the last year to a year and a half in a way that feels as un America maxing as possible. Jeffersalina, any thoughts before we move to the next questions?

Jeff Ding (57:18)

Yeah, I think the race question, I think to me it's tough because you almost have to imagine there's a counterfactual world with a different great power that is the same race as the US in this current moment. Would we still do a policy like the bad replant policy? Maybe. There's a lot of other reasons why that policy would still get implemented. John Dower Dover did this book on World War II and how the US conducted the war in the Pacific differently than they conducted the war against Germany. And there you have a counterfactual, you have something to compare it to, to tease out the role of race. And so he was able to identify sort of a racial component to that, that conduct of the war in the Pacific. I think it's harder to do that in this current context because there isn't that counterfactual scenario to compare it against.

Mika Ouyang (58:12)

At the same time, the history of anti Chinese Racism in the United States is deep. There's a reason why the Supreme Court case that was heard this week was brought by the plaintiff Wong Kim Ark, because the anti Chinese racism. Chinese, it's the only nationality that has been barred by statute from becoming naturalized citizens. When my mom's family moved to California in 1960, they were prohibited from buying a house because of racially restrictive covenants that would not allow them to sell to people of Chinese origin. Right. There were very similar kinds of racism against Chinese in the west that occurred against African Americans in the South. And so it's not like this is a new phenomenon either. But it is certainly the case that it exacerbates what is already suspicion. And it makes it harder to draw technical, rational, useful economic lines when there is so much underlying fear and it has such a deep history.

Kat Duffy (59:11)

All right, we started late, so I'm going to land this plane in five minutes, if that's okay. Okay, we have. I want to get to the back because there's so many people back there. Let me do these two on this side. I'm trying to be equitable. So right side, back, middle those two questions.

Jeff Ding (59:29)

I'm curious about the role of Congress in this debate, especially in the post doge context, what we were talking about. Well, you all were talking about funding, and I feel like, and I guess this is more widely perceived as a more hawkish Congress versus the executive, which is a little bit more amenable to. To making a deal and securing concessions. So I'm curious about what you all perceive as Congress's role in this debate.

Sam Sachs (60:05)

Hi. Thank you. I'm a graduate student at Georgetown looking into the use of AI in defense analysis. This week in my class, we talked about how only 2% of trade deals even mention the word AI. We talked about today the need for international standards on a technical level. What do you think those protections will look like? And do you think there's appetite internationally for that kind of discussion to get started, even not just between the US and China, but also between the US and the eu? Thank you.

Kat Duffy (60:37)

Okay, let's get quick answers to both of those. Who wants to take it? Anybody know what Congress should be doing, even if they're maybe not going to do it?

Mika Ouyang (60:48)

Just since I spent the beginning part of my career there, the first thing Congress needs to do is understand the technology. And there are not enough efforts to help them do that. And this is something I spent a lot of time as a staffer trying to help them understand really complicated technical things. They don't have A lot of help, a lot of the help that they get is driven by self interested commercial parties. There is not enough funding of researchers who can actually help inform Congress, who actually can put that language or put that analysis in terms that they understand, that can help them understand what functional oversight mechanisms are. I think there's a real risk about Congress getting it very, very wrong, which we saw happening in some of the export control supercomputer debates or getting sort of behind the curve on it if they're not really well educated. And that's not a smack against Congress. They come from many other parts of the world like we do not elect AI engineers to Congress. They don't make enough money to be like in Congress to be AI engineers. Why would you do that? But so we can't expect them to know how this technology is going to work. And it's not really fair to ask somebody who is like a pediatrician or a car salesman to understand this complicated technology without help. So like we have a lot of work to do.

Kat Duffy (62:12)

And when we think about things like trade agreements or incorporating AI into, not into non AI specific agreements, any thoughts there on the appetite for it and also the appropriateness of it integrating AI into like Internet, like trade agreements or other international agreements?

Selena (62:32)

I think on the question of like international standards or some kind of US China agreement on AI, I think there are quite a few areas and it kind of depends on. And also Chinese companies are very active in, you know, ISO and these international organizations. I think one of like some of the problems associated with those are like the problem of verification, like how exactly do you verify if either parties will adhere to, to the agreement. So I think when we talk about international standards, some of the things that already have a pre existing appetite that both countries have a pre existing appetite for would include you know, things for instance, like hey, like some level of transparency from companies that I think people tried to do at like the Bletchley park and then Seoul when they talked about like AI safety frameworks, whether there could be some way of trying to move some of those consensus areas into you know, international standards. And then also with the eu, there is the whole EU AI act. And I think a lot of Chinese companies are figuring out how exactly they want to adhere to parts of that in the code of practice. I think it's still tbd, but I think right now, because the US retreat probably will see some kind of movement from Chinese companies in trying to adhere to the code of practice and that kind of percolating back domestically into the standards that they have at home. But I'll just say that actually both sides, both ecosystems have both converging areas of risk perceptions, but also very different priorities in general. When speaking to Chinese AI companies, they're a lot more concerned about content safety and following government regulations at home, and less so on what American Frontier Labs are way more concerned about R and D uplift and Bio or cbrn. So I think there are some areas that they could converge and talk about just standards wise, even among companies themselves, without an international org. But there are also, I think, very different preconceptions on what are the standards of safety or security or governance that they want to aspire towards.

Kat Duffy (64:43)

Fantastic. And with that, I think we're gonna have to wrap. I feel like what we've heard from our panel is that the rivalry frame is real and legitimate in some respects, but that there are also costs to it, and that the perception of that frame as the sort of driving perception actually has its own negative externalities and costs that we should be watching out for. And I also think what's been really interesting for me, at least from the panel is hearing that we could be approaching competition and rivalry by really competing to be our best selves as well. I mean, who wants to be Tonya Harding when you could be Simone Biles, right? So I think our next session with Alondra and Henry is really going to lead us into what an affirmative vision for the United States could be and what that type of leadership would look like. And so with that, I want to thank my panelists and I will say on a personal note, with moderators privilege, I spent three years living in Cuba representing the US government under constant surveillance by an authoritarian state, being called a terrorist on the national news and a spy and a subversive and a contra revolutionaria and all these things. And if there is anything that living inside of an authoritarian state taught me, even though all of my work was directly with political dissidents, it is how nuanced and how disparate feelings were among the human beings that were sitting inside of that country and how binary and how reductive that conversation became when it became about the nation state powers and who controlled that narrative. And so with that, I also really want to thank our hosts for today for putting together an entire conference around the conversation that we're not having. Because so often the conversation that we are not having in the foreign affairs space is exactly the conversation that the human being sitting in any of our countries would be having with each other if they could. And so with that. I want to thank my panelists, thank all of you, thank our hosts, and so excited to welcome Alondra and her to the stage.

Mika Ouyang (67:23)

With VRBoCare.

Kat Duffy (67:25)

Help is always ready before, during and after your stay.

Mika Ouyang (67:30)

We've planned for the plot twists, so support is always available because a great trip starts with peace of mind.

Selena (67:39)

You know what they say. Early bird gets the ultimate vacation home. Book early and save over 150 $20 with VRBO because early gets you closer to the action, whether it's waves lapping at the shore or snoozing in a hammock that overlooks. Well, whatever you want it to so you can all enjoy the payoff come summer with VRBO's early booking deals. Rise and shine average savings $141 select homes only.