alt Hacker NewsI’d hope this isn’t a goal post move - an open math problem of any sort being solved by a language model is absolute science fiction.
Replies
It is not. You're operating under the assumption that all open math problems are difficult and novel.
This particular problem was about improving the lower bound for a function tracking a property of hypergraphs (undirected graphs where edges can contain more than two vertices).
Both constructing hypergraphs (sets) and lower bounds are very regular, chore type tasks that are common in maths. In other words, there's plenty of this type of proof in the training data.
LLMs kind of construct proofs all the time, every time they write a program. Because every program has a corresponding proof. It doesn't mean they're reasoning about them, but they do construct proofs.
This isn't science fiction. But it's nice that the LLMs solved something for once.
That's been achieved already with a few Erdös problems, though those tended to be ambiguously stated in a way that made them less obviously compelling to humans. This problem is obscure, even the linked writeup admits that perhaps ~10 mathematicians worldwide are genuinely familiar with it. But it's not unfeasibly hard for a few weeks' or months' work by a human mathematician.