“One of the reasons we were successful in the old days with arms control is because we came to an understanding that you couldn't win a nuclear arms race. … So the only way, if I were to be less skeptical, is for us to genuinely believe on both sides that the pursuit of dominance is unachievable and undesirable. … It isn't the most important thing in the world to be number one here.”
The Spillover · October 2026
Context
Asked whether the US and China can find common ground on AI governance, Steinberg said he's skeptical. Too many people on both sides think being first, perhaps through recursive self-improvement, would bring a permanent advantage, and as long as they do, agreement is out of reach.
Transcript
…the president is trying to say about how is China an enemy? Isn't it an adversary? Is it trying to do us harm? And we have to do everything we can to prevent that from happening. Or is this a country that has differences, but it's all kind of a transaction and we can work our way through it.
…So global cooperation on AI rolls of the road, AI governance, seems like something that's pretty important for both places. I know they have started an AI dialogue between the U.S. and China. Are you hopeful at all on that? Do you think we can find any common ground on AI governance with China?
…winning in that sense becomes taking all the chips on the table. And one of the reasons we were successful in the old days with arms control is because we came to an understanding that you couldn't win a nuclear arms race, that nuclear war could never be one never should be fought, that trying to get ahead or getting a first strike dominance was just not achievable and risky to try to do. So the only way, if I were to be less skeptical, is for us to genuinely believe on both sides that the pursuit of dominance is unachievable and undesirable. And I do think there are some voices that are trying to argue that now, that it's a false hope and that pursuing it is dangerous, that it leads to instability. But it will take both sides to really genuinely be willing to say that and say we're not going to try. It isn't the most important thing in the world to be number one here and that we need to find ways to focus on the risks of basically…
Do you think it's going to take something bad happening to Get people at the table to think about kind of a nuclear arms deal version of AI? Do we have to have some sort of AI-related bio event or a cyber event that is much more significant, takes out the electricity grid for a country or financial…
It's a good point. And it may be that one is necessary and hope one that we survive. I mean, we all recall, some of us are old enough to recall it, you know, that a pivotal point in the U.S. approach for President Kennedy was after the Cuban Missile Crisis in his famous America’s Friends speech…
Oh, see, that's a problem, too. I was going to say that one thing that would give me some measure of hope on the U.S. side is that if you read opinion polls, the American public is increasingly anxious about AI, whether we're talking about electricity costs or water or jobs, especially for young…
More from The Spillover
