Dwarkesh Patel: observation

Transcript context

…I don't know. That feels orthogonal because getting a gold in the IMO feels a lot more like being really, really good at Go or Chess. Those feel analogous. It's super creative. I don't know chess as well as the people who are into it, but everything that I hear from them, the sort of moves that are made and choices have all of the air of creativity. I think as soon as they started generating artwork, then everyone else could appreciate, “Oh, there's something that deserves to be called creative here.” I don't know how it would look when people get them to be getting golds at the IMO but I imagine it's something that looks a little bit like how AlphaGo is trained, where you have it play with itself a whole bunch. Math lends itself to synthetic data in the ways that a lot of other domains don't. You could have it produce a lot of proofs in a proof checking language like Lean, for example, and just train on a whole bunch of those. And ask, is this a valid proof? Is this not a valid proof? And then counterbalance that with English written versions of something. I imagine what it looks like once you get something that is solving these IMO level things, is one of two things. Either it writes a very good proof that you feel is unmotivated, because anyone who reads math papers has this feeling that there are two types. There are the ones where you morally understand why the result should be true and then there are the ones where you're like, “I can follow the steps. Why would you have come up with that? I don't know. But I guess that shows that the result is true.” And you're left wanting something a little bit more. And so you could imagine if it produces that to get a gold in the IMO, is that the same kind of ability as what is required to replace jobs? Not really. The impediments between where it is now and replacing jobs feels like a whole different set of things like having a context window that is longer than some small thing such that you can make connections over long periods of time and build relationships and understand where someone's coming from and the actual problem solving part of it. It's a sign that it would be a more helpful tool, but in the same way that Mathematica can help you solve math problems much more effectively. Tell me why I should be less amazed by it or maybe put it in a different context but the reason I would be very impressed is… With chess, obviously this is not all the chess programs are doing, but there's a level of research you can do to narrow down the possibilities. And more importantly, in the math example, it seems that with some of the examples you've listed on your channel, the ability to solve the problem is so dependent on coming up with the right abstraction to think about it, coming up with ways of thinking about it that are not evident in the problem itself or in any other problem in any other test, that seems different from just a chess game where you don't have to think about what is the largest structure of this chess game in the same way as you do with the IMO problem. I think you should ask people who know a lot about Go and Chess and I'd be curious to hear their opinions on it because I imagine what they would say is, if you're going to be as good at Go as AlphaGo is you’re also not doing tree search, at least exclusively. It's not dependent on that, because you get this combinatorial explosion, which is why people thought that game would be so much harder for so much longer. There sort of has to be something like a higher level structure in their understanding. Don't get me wrong, I anticipate being very impressed when you get AIs that can solve these IMO problems, because you're absolutely right, there's a level of creativity involved. The only claim I'm making is that being able to do that feels distinct from the impediments between where we are now and the AIs take over all of our jobs or something. It seems like it's going to be another one of those boxes that's this historic moment analogous to chess and Go, more so than it's going to be analogous to the Industrial Revolution.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.