Minichan

Topic: Idk why I was bored I guess so I looked at OpenAI’s math proofs

Live&Lemon started this discussion #138,847

Idk how I feel about it. On the one hand, it would be good if we could progress mathematics further. I’m not a mathematician, but I can kind of see what mathematicians were talking about now. A lot of the proofs are counter examples. So it’s sort of like if someone said "hey I don’t know if this thing is true or false" and then somebody else gave an example where it doesn’t work. So you know that it’s false now, but the fact that you know that it’s false doesn’t mean it’s clear what you do with that information. For example right now I’m looking at a PDF on GitHub that is a counter example to the critical number Arnold bound. It’s a counter example which disproves the unrestricted critical-number and rational cup-length forms of the Arnold conjecture. So I know that there isn’t an unrestricted critical-number and rational cup length form of the Arnold conjecture. But like… what do I do with that information exactly? If nobody ever was operating under the assumption that it was either true or false before, and now we know that it’s false, how do we apply this information?

(Edited 17 seconds later.)

Live&Lemon (OP) double-posted this , 16 minutes later[^] [v] #1,468,346

I kinda get the point that mathematicians were making that it sort of just seems like creating work for mathematicians, because they brute forced a bunch of counter examples to various problems, but the fact that the counter example is valid didn’t require explaining or coming up with any new techniques for finding solutions. So it’s just a bunch of stuff to for humans to verify that the lean and the natural language proofs match (some of them don’t because the natural language proofs are AI generated translations based on the AI generated lean code).

https://github.com/openai/math/tree/main/preprints

I hope something useful comes out of this, but I find myself kinda doubting it, especially as I was scrolling online I was seeing people saying they’re reconsidering not doing PhDs because of this. But nobody who doesn’t have a PhD is ever going to look at any of this and do anything useful with it.

Live&Lemon (OP) triple-posted this , 1 minute later, 18 minutes after the original post[^] [v] #1,468,347

(PDFs are binary formatted files not code so you have to view them as raw on a desktop in order to see them or download them if you’re not on a desktop)
:

Please familiarise yourself with the rules and markup syntax before posting.