Georgetown CSET Executive Director @hlntnr explains why AI-enabled bio attacks could become a rare point of common ground between America and China: "The right counterargument from a US perspective would be, okay, but maybe Chinese models will lead to bio attacks outside China that then affect China. Even if you have perfect censorship and perfect surveillance of all your citizens, maybe you can't stop someone outside China from doing this." "Another one I've been thinking about is incident reporting after Hugging Face. There's a lot of questions around what happened here, and what conclusions we should draw." "That's the kind of thing where I think it's very unlikely we'll get a handshake deal of we'll do X if you do Y, just because the trust isn't there." "It needs to be something more like, hey, we think it would be a good idea if both of us do X, and we can just do it. It's cost-free for the US to say, look, we can start sharing some incidents with you. Trying to get them to come along, rather than trying to get some kind of hard commitment." @CSETGeorgetown

Georgetown CSET Executive Director @hlntnr reveals why China sees AI companions as a threat to the Communist Party even when they never criticize it: "In China, everything you do online is tied back to your SIM card, which is tied back to your national ID card. They can look at anything anyone is doing online and trace back to an individual person." "There's one thing, does the AI tell you the Communist Party is bad. But there's a different thing, does AI cause social effects that have knock on dissatisfaction with the Communist Party. Their most recent regulation was actually on AI companions. They called it human-like AI. Read the f---ing regulation." "The motivation there is less AI companions will tell their users that the CCP is bad, and more like if you have kids committing suicide, if you have these broader negative effects on society, is that bad for their hold on power. That's the primary lens they're looking at all this through." @CSETGeorgetown

Georgetown CSET Executive Director @hlntnr explains why China thinks it takes AI safety more seriously than America does: "The problem is that both sides take it seriously in their own way. Sometimes in the US there's a sense that China is kind of a little bit behind us on this path to taking AI more seriously. But actually they sort of have the flip side, the same view of us." "There was this really interesting piece that came out a week or two ago. The title was 'Anthropic Has American Disease.' It was criticizing the US government for not being serious on AI regulation." "From the Chinese perspective, China has all these regulations, technical standards, all this stuff going on with their generative AI sector, and the US is really unserious because everything we're doing is voluntary." "So both sides think that the other is not taking it seriously. For me it's less like will China be serious enough, and more like can they identify some specific concrete issue where they don't just talk past each other?" @CSETGeorgetown

Georgetown CSET Executive Director @hlntnr argues China may not actually want to race America to AGI: "It's really not clear to me that they are. They think of themselves as in competition with the US on AI. They wanna be leading the world of AI. There's a specific version of racing to recursive self-improvement so that you're the first one to get to takeoff so that you have a decisive strategic advantage forever. I really don't think that is a prominent set of ideas in China." "Some people are like, well, if they only understood how powerful AGI will be, then obviously they would have to race. Which I think is not right, given how much they care about being in control." "The Chinese Communist Party, their whole deal is staying in control, keeping society stable. If they thought there was a serious risk that they would lose control of AI, then that would not feel appealing to them." "It often feels like there's a real nihilism behind the, oh well, we could try and prevent them, but we can't because China. That just seems so defeatist, giving up so early in the story." @CSETGeorgetown

Georgetown CSET Executive Director @hlntnr reveals why the term "AGI" is becoming useless just as we're getting closer to it: "AGI as a term is kind of a fuzzy cloud. When it was a fuzzy cloud that was far away, you could say oh AGI and gesture towards it, and that was a useful thing to say. But we're now inside the fuzzy cloud of different things you might mean by AGI. It's not really very helpful to say is this AGI or is it not. It all depends on how you define it." "It doesn't feel useful to talk about in the abstract is Astra AGI or not. People are trying to get at some tipping point. But people mean quite different tipping points." "Are you talking about AI that can tip you into fully automated AI R&D, recursive self-improvement, intelligence explosion? Or AI that is a drop-in remote worker and can wipe out large chunks of the workforce. That's a different tipping point." "An approach to jargon I keep coming back to is, if you can just say a short phrase instead of some jargon term, that's almost always better. If you can say super duper powerful AI, no really super powerful. Or if you mean AI that can kick off RSI, say that. The fact we have to have a whole discourse cycle about it shows this term is not very helpful." @CSETGeorgetown