Pace AI progress with concrete pause triggers, not a growth cap

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

1. 1:45:59

   Yeah, that's not a very comforting description of where we are. Yeah. When it comes to all these ideas about pacing, and the paper didn't really quite go there, but I assume that you probably had to have some conversations with your co-authors around, okay, we're all excited about pacing. What does that look like? What are you hearing from, and what notions of Pitt seem promising? Yeah, there's been a lot of discussion about pacing.

2. 1:46:34

   We decided not to have something more concrete in the paper. It's just like a big trade-off in terms of time, basically. So yeah, definitely I think it'd be very valuable to have a big multi-author thing. I'm like, this is the plan to pace the frontier. But okay, doing that might take three years instead of six-ish months. So yeah, although, yeah, I do think there are a lot of plans to just individually sort of follow up on that.

3. 1:46:59

   Yeah, and just like what should be paced. So I think there are a couple of ways of thinking about it. I have sort of an unconventional view. So maybe I'll describe like the conventional view, it seems, which is, yeah, paced in the frontier. What does that mean? You sort of put a cap on the rate of AI progress, whether that's through compute, whether that's through just doing research, maybe it's through data.

4. 1:47:17

   I don't really know. But just like a cap on the rate of capability is improvement. That's enforced. So I think that's pretty intuitive for people. And yeah, I think that could potentially be a good idea. I don't really know what the number should be or whatever, but yeah, worth what I think trying to figure out. Maybe my more unconventional idea of pacing is much more related to classic risk management.

5. 1:47:41

   So in some sense, even though OpenAI hasn't committed to a cap on its rate of progress, it is still in some sense pacing the frontier. because at least according to its recent commitments, it's like, okay, if we see a super crazy loss of control event, we're going to pause our training and try to figure out what went down. So even though it's not sort of phrased explicitly as, we're going to pace to 2x algorithmic improvement or whatever, it effectively leads to a slowdown because as capabilities improve, you're going to encounter more of these sorts of loss of control or other types of problems.

6. 1:48:16

   And if you sort of follow your commitment, you're going to pause to try to figure out what went down. And the reason it sort of appeals to me is, well, yeah, maybe for two reasons. One is like, it seems much more tractable to get agreement on these sorts of concrete risk management type, you know, ideas than just like a kind of very abstract number, you know, a cap on the rate of AI progress. And number two, I think this sort of stuff actually gets much more to the core of the AI problem, which is we have some idea at least now of the problems that we want to avoid and indicators that lead to those problems.

7. 1:48:52

   When we see those indicators, we should probably pause and try to figure out what went down before going forward again. So yeah, that's the type of... I guess the second seems much more neglected right now, at least in the current discourse, and they're trying to figure out, yeah, what to say and write on that. It jumps out to me that Anthropic basically had a version of that in their responsible scaling policy, and then they retracted it, right?

8. 1:49:15

   They used to have a whole bunch of if-then commitments, that if we are at this level of capability and we can't do this, then we have to basically stop training. And now my sense is basically like they're deciding as they go what is prudent to do. And does that, do you see a way back? I mean, it is striking that, you know, in a way, OpenAI maybe is even doing a little more of, like, concrete if then than Anthropic at this point.

9. 1:49:43

   That may be a little bit of an overcorrection, but, like, certainly it used to be Anthropic's thing, and now it's maybe a little bit more OpenAI's thing. I don't know. Yeah, yeah, yeah. Yeah, I mean, I'm, like, cautiously optimistic that there's a way back. The way back, I think, is through embedded auditors. I think a big reason why, at least in version 3 of the RSP, why Anthropic removed that sort of explicit commitment in version three of their RSP is concerns about like race dynamics, basically, that like if they pause, then from their lights, a less cautious competitor would go ahead. But a good embedded auditing system should address that to some extent, because it allows you to verify to the public and to others that, or maybe not to a high level of confidence, but like, you know, assure to some level of confidence

10. 1:50:26

    that you're actually following through on that commitment, which helps in other people or other labs following through on the commitment as as well so yeah maybe we'll see something like that i think i yeah how does this work ideally like it does seem like to me there has to be some sort of minimum common minimum safety standard through some sort of like body like what is that body is it a finra is it like a mutual is it like something like nist i'm not really sure but i think it does have to come from some sort of common body and be promulgated to to its all to all its members. And then, yeah, and then he's involved somehow and making sure that members follow, follow through. Well, we now have a one pager after an afternoon at the White House, at least. So we've got that much. Come for us. It's a step, no doubt. This mutual idea. I've