Fields Medalist Alessio Figalli Speaks Out About AI Mathematicians

My last questions have to do with this announcement. Two days ago, OpenAI announced that its model had constructed a rigorous mathematical existence proof for four-dimensional Yang-Mills theory, and that's one half of a Millennium problem. And David Gross, who won the Nobel Prize for work with QCD. My question is twofold. You just mentioned that there needs to be follow-up by experts, but my question is there's not much transparency going on. So your fellow mathematicians are demanding that OpenAI release the proofs immediately rather than giving privileged access to people like David Gross just because he has a Nobel. Yeah, that's a very good question. I think, I don't know what is the rationale behind that. I haven't seen the proofs. Nobody has besides these four people. Yeah, exactly. So I haven't seen the proofs. I haven't seen anything. I mean, there have been, I think, consultations also with the advisory board that they have now, which is made of some mathematicians. I don't know if there was an idea. The Altmutter Island... I don't know what was the status there, right? So if they wanted to, what was the rationale? I mean, that's the part that perhaps one should ask OpenAI. For me, anyhow, giving up, I mean, there was also a bit of that when they tried to do with endless problems, to have people check. I think the problem now is more like you don't know what comes next, right? So it's like every month, every two months, I don't know, every week, new problems come and are solved. There's a lot of uncertainty. To be honest, I don't know. Perhaps this is a moment, in the sense that once the model has reached this super-human capability and potentially already reached, I don't know until when OpenAI will have a motivation to keep solving just math problems. I mean, I don't even see a clear purpose there to keep using these models just to solve more math. I mean, the production anyhow is increasing exponentially overall, just because people have AI at their disposal. I would expect that in a couple of years we would be more of a steady state where, you know, we will know more how things evolve. I think on our side, what we can do is just to say, OK, we choose the problem that really matters to us. We work on them because we want to understand them in our way. We go on and we make sure that our students, you know, get the training and get everything they need to keep going, to be supported. They see the value of understanding. They're not damaged by spending time to understand. You know, that's, I think, the most important thing. So we don't want them to feel that they need necessarily to use AI themselves to speed up. We want them to have the time to learn at their pace because that's the only way to train the future generation.