I agree that I think it’s wading into tricky territory, because worries over China’s competitiveness can fuel further development as well as safety, but it seems to me that choosing very particular interventions as follow-ups to that framing can potentially reduce that risk.
Location verification mechanisms are a great example, where the framing can even just be “This is a strategic technology and we know other countries are violating rules we’ve placed around exports. Our law should be respected and this is one way to achieve that.”
Again, I understand the worry, but I do think with careful framing it can be safety supporting.
Yeah, I don’t think all Nat Sec stuff is bad. Competition and rivalry here is inevitable to some degree, and we really do need Nat Sec people to help manage the US end of it, especially as they are also the people who know most about making treaties and deals between rivals.
We also want to avoid a conventional war between the US and China (plausibly over Taiwan and TSMC). Although I guess there is an argument that X-risk considerations should dominate that, but I think I go with commonsense over longtermist philosophical argument on that one, and probably such a war is very bad for X-risk anyway. We also want better US-China relations during the AI race. Both of these goals are somewhat threatened by the US trying to aggressively contain China. EDIT: I should say, they are threatened by the US trying to aggressively contain China’s AI development in the short term. Obviously, taking up a more aggressive posture on defending Taiwan itself might deter China rather than starting a war (maybe). And of course, once transformative AI is here, China being significantly ahead in the race makes China occupying Taiwan more likely, though whether it raises the risk of a great power US-China war which is what we really should care most about is much less clear.
Ah okay, I better understand what you mean when you say Natsec now. On the China front, do you think there’s any advantage at all to delaying their capacity for developing AI? To put that another way, is there any degree of increased risk of a US-China conflict that you’d be willing to accept for delaying China’s AI development?
As a small point, the world in which China has developed TAI before we have, and has taken back Taiwan, doesn’t seem stable at all to me. There is a sense in which what will happen is less clear just by virtue of the world now having TAI, but it seems fairly clear to me that China taking Taiwan would raise tensions, and that it would be arguably worse in such a time because the US would already have to be contending with the fact it lost the AI race. So I don’t think you can dismiss the risks such a situation would pose so quickly, and it’s not clear to me that even if opposing war between the two powers was your only objective that it would be safer to not take any protective action now.
On the last point, I wonder if your definition of nat sec is more broad than mine? I think that most of the examples you cite seem to much more squarely fit into reducing the spread of communism, which I see as fair distinct from what I traditionally view as nat sec policy. It feels like the inherently global and non-isolationist actions taken to prevent the spread of communism stand as a partially opposed approach even, because nat sec policy often seems to involve walking back the US’s connection to other places in the world. At the very least, I think that current nat sec people would look fairly distinct from those pushing these past policies.
Thanks for the multiple sources though, I think some of the citations there do seem to paint a very negative picture of US actions in those places, though I think it’s only properly clear in the case of the Indonesia mass murders and seems to be a bit more uncertain elsewhere. Do you know any good book or singular resource that really digs into these cases (and those similar) that you would recommend?
There is plausibly some advantage from delay yes. For one thing even if you don’t have any preference for which side wins the race, making the gap larger plausibly means the leading country can be more cautious because their lead is bigger, and right now the US is in the lead. For another thing, if you absolutely forced me to choose, I’d say I’d rather the US won the race than China did (undecided whether the US winning is better/worse than a multipolar world with 2 winners). It’s true that the US has a much worse record in terms of invading other places and otherthrowing the governments than China, but China has not had anything like the US’s international clout until recently, so it’s unclear how predictive past behaviour on China’s part is of future behaviour. And on the other hand, China is, while apparently well-governed in many ways, very authoritarian, which I think is bad. (Although the US may be about to go less than fully democratic, it would have to fall far to be as authoritarian as China, though it does imprison a much higher % of its population than China I think.) I generally would not want to see authoritarianism win out in some general sense, even if China itself might be a more restrained actor than the US in many ways.
Yeah, maybe I am using the word Nat Sec wrong, but my sense is that US intelligence agencies were involved in at least some of the history I was mentioning. I am very much not an expert on that history, but I recall Matt Yglesias recommending this (which I haven’t read to be clear): https://en.wikipedia.org/wiki/The_Jakarta_Method I don’t think Yglesias is particularly at all expert of particularly reliable on this stuff either, but I do think he generally has a fairly (civic) nationalistic pro-US point of view, so if the book persuaded even him that the US did a lot of bad stuff in Indonesia and elsewhere during The Cold War, it probably marshals quite a lot of evidence for that conclusion, and probably isn’t too partisanly tankie.
I agree that I think it’s wading into tricky territory, because worries over China’s competitiveness can fuel further development as well as safety, but it seems to me that choosing very particular interventions as follow-ups to that framing can potentially reduce that risk.
Location verification mechanisms are a great example, where the framing can even just be “This is a strategic technology and we know other countries are violating rules we’ve placed around exports. Our law should be respected and this is one way to achieve that.”
Again, I understand the worry, but I do think with careful framing it can be safety supporting.
Yeah, I don’t think all Nat Sec stuff is bad. Competition and rivalry here is inevitable to some degree, and we really do need Nat Sec people to help manage the US end of it, especially as they are also the people who know most about making treaties and deals between rivals.
We also want to avoid a conventional war between the US and China (plausibly over Taiwan and TSMC). Although I guess there is an argument that X-risk considerations should dominate that, but I think I go with commonsense over longtermist philosophical argument on that one, and probably such a war is very bad for X-risk anyway. We also want better US-China relations during the AI race. Both of these goals are somewhat threatened by the US trying to aggressively contain China. EDIT: I should say, they are threatened by the US trying to aggressively contain China’s AI development in the short term. Obviously, taking up a more aggressive posture on defending Taiwan itself might deter China rather than starting a war (maybe). And of course, once transformative AI is here, China being significantly ahead in the race makes China occupying Taiwan more likely, though whether it raises the risk of a great power US-China war which is what we really should care most about is much less clear.
More over there is also just a long record of the US doing really, really bad shit for nat sec reasons that makes me feel nervous about aligning with nat sec people even though I think it is probably sometimes necessary and I’m not saying no one should ever fund a nat sec orientated thing: like when they gave out death lists of communists to the Indonesian government and then 5.-1 million people were murdered (https://en.wikipedia.org/wiki/Indonesian_mass_killings_of_1965%E2%80%9366), or the other time they backed a genocide in Indonesia (https://en.wikipedia.org/wiki/East_Timor_genocide), or the time they backed a genocide in Guatemala where “national security” mostly meant the interests of US fruit companies https://en.wikipedia.org/wiki/Guatemalan_genocide or their complicity in genocide in Bangladesh (https://en.wikipedia.org/wiki/Bangladesh_genocide#Operation_Searchlight). Things have gotten a bit better since the Cold War, but look what is happening in Gaza right now.
Ah okay, I better understand what you mean when you say Natsec now. On the China front, do you think there’s any advantage at all to delaying their capacity for developing AI? To put that another way, is there any degree of increased risk of a US-China conflict that you’d be willing to accept for delaying China’s AI development?
As a small point, the world in which China has developed TAI before we have, and has taken back Taiwan, doesn’t seem stable at all to me. There is a sense in which what will happen is less clear just by virtue of the world now having TAI, but it seems fairly clear to me that China taking Taiwan would raise tensions, and that it would be arguably worse in such a time because the US would already have to be contending with the fact it lost the AI race. So I don’t think you can dismiss the risks such a situation would pose so quickly, and it’s not clear to me that even if opposing war between the two powers was your only objective that it would be safer to not take any protective action now.
On the last point, I wonder if your definition of nat sec is more broad than mine? I think that most of the examples you cite seem to much more squarely fit into reducing the spread of communism, which I see as fair distinct from what I traditionally view as nat sec policy. It feels like the inherently global and non-isolationist actions taken to prevent the spread of communism stand as a partially opposed approach even, because nat sec policy often seems to involve walking back the US’s connection to other places in the world. At the very least, I think that current nat sec people would look fairly distinct from those pushing these past policies.
Thanks for the multiple sources though, I think some of the citations there do seem to paint a very negative picture of US actions in those places, though I think it’s only properly clear in the case of the Indonesia mass murders and seems to be a bit more uncertain elsewhere. Do you know any good book or singular resource that really digs into these cases (and those similar) that you would recommend?
There is plausibly some advantage from delay yes. For one thing even if you don’t have any preference for which side wins the race, making the gap larger plausibly means the leading country can be more cautious because their lead is bigger, and right now the US is in the lead. For another thing, if you absolutely forced me to choose, I’d say I’d rather the US won the race than China did (undecided whether the US winning is better/worse than a multipolar world with 2 winners). It’s true that the US has a much worse record in terms of invading other places and otherthrowing the governments than China, but China has not had anything like the US’s international clout until recently, so it’s unclear how predictive past behaviour on China’s part is of future behaviour. And on the other hand, China is, while apparently well-governed in many ways, very authoritarian, which I think is bad. (Although the US may be about to go less than fully democratic, it would have to fall far to be as authoritarian as China, though it does imprison a much higher % of its population than China I think.) I generally would not want to see authoritarianism win out in some general sense, even if China itself might be a more restrained actor than the US in many ways.
Yeah, maybe I am using the word Nat Sec wrong, but my sense is that US intelligence agencies were involved in at least some of the history I was mentioning. I am very much not an expert on that history, but I recall Matt Yglesias recommending this (which I haven’t read to be clear): https://en.wikipedia.org/wiki/The_Jakarta_Method I don’t think Yglesias is particularly at all expert of particularly reliable on this stuff either, but I do think he generally has a fairly (civic) nationalistic pro-US point of view, so if the book persuaded even him that the US did a lot of bad stuff in Indonesia and elsewhere during The Cold War, it probably marshals quite a lot of evidence for that conclusion, and probably isn’t too partisanly tankie.