Should governments now impose a strong pause or significant slowdown on frontier AI development?

“I do think we should be skeptical of those arguments, but at the same time, yeah, I do think that this is an additional reason why if we're going to pause, we should, you know, do a deal with China, make sure they're pausing as well.”

“and I guess, like, in practice, it seems like, to my knowledge, we're nowhere close currently to, like, being confident that, our AIs are aligned. and so I, I think near term, at least, the thing to do is to work on not building it. Which is to say, like, computer, stopping it, pausing it, slowing it, ideally stopping it,”

“a pause would be great, but I actually think the framing is not quite right. Like, we should freeze development where it is, and then evaluate which direction we want it to go”

“the retort to that is like, well, like why is it assumed that we want to go any further? Like maybe we in fact don't want to go any further and we should like bring the whole thing to a grinding halt and then kind of reassess whether this like approach of doing, you know, aligning our AIs using weaker”

“I didn't think that that was the case. I was just like, oh, it seems like we should just stop as soon as possible because this whole thing seems dangerous. Whereas now, I'm like, oh, it seems like we're actually really, really getting close to the thing and we have so much evidence that we're getting close to the thing”

“if in a year everything's chill and nothing's happened and someone asked me, would I still endorse saying we should pause now, I think I would say that I endorse it. Like, I think like just the states seem high enough now that like we clearly just should chill out for a bit.”

“Then I'm more optimistic. I'm like, yes, then we will take, you know, GPT-6, GPT-7, whatever the most advanced AI models we have, and we will apply them to the task of helping us figure out how these neural networks work.”

“Pause is a simple policy and it's really what should precede any next step. You don't make the problem worse before you figure out your plan and do it.”

“The actual regulatory thing we need to do today is cut off the compute, slow down these labs fully, and I don't, I don't care about China here.”

“you don't want to race ahead to beat China and then you realize that, you know, you've ruined democracy in the US. And so now, you know, you try to avoid extreme power concentration in the hands of China, but you've just got extreme power concentration in the hands of a few people in the US.”

“I think that we should freeze frontier development right now, and we should negotiate a bilateral agreement with China to make that binding, and we should establish verification, both on a technical level, and then you could use things like third-party auditors and compute registration.”

“It's very hard to monitor OpenAI and an anthropic pacing, you know, it's, but it's very easy to monitor if they have paused, like if certain activities have ceased, you know, whether they ever resume again or not.”

“it's actually way, way harder to just slow down and pace or whatever that means than to pause.”

“for company coups, I think there's a very strong case that slowing down would help. it would allow the rest of the world to understand what these companies are building. It would allow multiple companies to kind of, and multiple countries to catch up to the frontier. and so I think the case there is very robust.”

“if you stop and you hit the pause button through some regulatory mechanism, you effectively allow six companies to over-earn. You lock it in”

“if you think about how fast this happened, this is like three weeks. And you have this solution where the alternative was hit the pause on trillions of dollars of GDP so that we could sort this out in a multinational body. The alternative was crazy.”

“we cannot lose our lead to the Chinese. ... it would be very different if they had the lead in AI rather than us, both from a capability point of view, how they would be using it, but more importantly, their willingness to discuss it.”

“And the reason that, if there were no downside... To regulating AI and stopping it in its tracks and turning it off, I'd probably be on board with you guys because then it's just a research project for research money back.”

“They, nothing would change other than we're losing time, foreign governments are getting better, and you're still not doing anything. You're, you're moving at a pace that is so behind the rate of AI. So all that would happen if we quote unquote slowed down is we would fritter away the time.”

“And I, I think the worst possible outcome is deceleration, so don't decel.”

“I didn't think that that was the case. I was just like, oh, it seems like we should just stop as soon as possible because this whole thing seems dangerous. Whereas now, I'm like, oh, it seems like we're actually really, really getting close to the thing and we have so much evidence that we're getting close to the thing”

“It's very hard to monitor OpenAI and an anthropic pacing, you know, it's, but it's very easy to monitor if they have paused, like if certain activities have ceased, you know, whether they ever resume again or not.”

“The actual regulatory thing we need to do today is cut off the compute, slow down these labs fully, and I don't, I don't care about China here.”

“I think that we should freeze frontier development right now, and we should negotiate a bilateral agreement with China to make that binding, and we should establish verification, both on a technical level, and then you could use things like third-party auditors and compute registration.”

“I do think we should be skeptical of those arguments, but at the same time, yeah, I do think that this is an additional reason why if we're going to pause, we should, you know, do a deal with China, make sure they're pausing as well.”

“and I guess, like, in practice, it seems like, to my knowledge, we're nowhere close currently to, like, being confident that, our AIs are aligned. and so I, I think near term, at least, the thing to do is to work on not building it. Which is to say, like, computer, stopping it, pausing it, slowing it, ideally stopping it,”

“the retort to that is like, well, like why is it assumed that we want to go any further? Like maybe we in fact don't want to go any further and we should like bring the whole thing to a grinding halt and then kind of reassess whether this like approach of doing, you know, aligning our AIs using weaker”

“if in a year everything's chill and nothing's happened and someone asked me, would I still endorse saying we should pause now, I think I would say that I endorse it. Like, I think like just the states seem high enough now that like we clearly just should chill out for a bit.”

“it's actually way, way harder to just slow down and pace or whatever that means than to pause.”

“a pause would be great, but I actually think the framing is not quite right. Like, we should freeze development where it is, and then evaluate which direction we want it to go”

“Then I'm more optimistic. I'm like, yes, then we will take, you know, GPT-6, GPT-7, whatever the most advanced AI models we have, and we will apply them to the task of helping us figure out how these neural networks work.”

“Pause is a simple policy and it's really what should precede any next step. You don't make the problem worse before you figure out your plan and do it.”

“And I, I think the worst possible outcome is deceleration, so don't decel.”

“if you stop and you hit the pause button through some regulatory mechanism, you effectively allow six companies to over-earn. You lock it in”

“And the reason that, if there were no downside... To regulating AI and stopping it in its tracks and turning it off, I'd probably be on board with you guys because then it's just a research project for research money back.”

“if you think about how fast this happened, this is like three weeks. And you have this solution where the alternative was hit the pause on trillions of dollars of GDP so that we could sort this out in a multinational body. The alternative was crazy.”