• 10 months ago
OpenAI CEO Sam Altman spoke at the World Economic Forum in Davos on Thursday and commented on the company's leadership drama and the concerns over Artificial Intelligence.
Transcript
00:00 At some point, you just have to laugh.
00:02 Like at some point, it just gets--
00:04 it's so ridiculous.
00:05 But I think-- I mean, I could point to all the obvious
00:09 lessons that you don't want to leave important--
00:13 you don't want important but not urgent problems out there
00:16 hanging.
00:16 And we had known that our board had gotten too small.
00:19 And we knew that we didn't have the level of experience
00:21 we needed.
00:22 But last year was such a wild year for us in so many ways
00:24 that we sort of just neglected it.
00:27 I think one more important thing, though,
00:31 is as the world gets closer to AGI, the stakes, the stress,
00:39 the level of tension, that's all going to go up.
00:43 And for us, this was a microcosm of it,
00:45 but probably not the most stressful experience
00:47 we ever face.
00:50 And one thing that I've sort of observed for a while
00:54 is every one step we take closer to very powerful AI,
01:00 everybody's character gets like plus 10 crazy points.
01:05 It's a very stressful thing.
01:07 And it should be, because we're trying
01:09 to be responsible about very high stakes.
01:11 And so I think that as--
01:15 I think one lesson is as we get--
01:19 we, the whole world, get closer to very powerful AI,
01:24 I expect more strange things.
01:27 And having a higher level of preparation, more resilience,
01:33 more time spent thinking about all of the strange ways
01:36 things can go wrong, that's really important.
01:39 Well, I don't think they're guaranteed to be wrong.
01:41 I think there's a spirit.
01:43 There's a part of it that's right,
01:44 which is this is a technology that is clearly very powerful
01:49 and that we don't know-- we cannot say with certainty
01:53 exactly what's going to happen.
01:54 And that's the case with all new major technological
01:57 revolutions.
01:58 But it's easy to imagine with this one
02:03 that it's going to have massive effects on the world
02:05 and that it could go very wrong.
02:09 The technological direction that we've
02:10 been trying to push it in is one that we think we can make safe.
02:15 And that includes a lot of things.
02:16 We believe in iterative deployment.
02:19 So we put this technology out into the world along the way
02:23 so people get used to it, so we have time as a society,
02:26 our institutions have time to have these discussions,
02:28 figure out how to regulate this, how to put some guardrails
02:31 in place.
02:32 If you look at the progress from GPT-3 to GPT-4
02:36 about how well it can align itself to a set of values,
02:40 we've made massive progress there.
02:41 Now, there's a harder question than the technical one,
02:44 which is who gets to decide what those values are.
02:46 And what the defaults are, what the bounds are,
02:48 how does it work in this country versus that country,
02:51 what am I allowed to do with it versus not.
02:54 So that's a big societal question, one of the biggest.

Recommended