llms and math discoveries!
llms have been on my mind while chatting with a few startups (see my previous post on how i feel about that) in some negative ways - but the recent jacobian conjecture counterexample is pretty cool! I took a lot of math courses during college so it's pretty sick to see one of the more annoying processes of math (slapping around tons of counterexamples until one sticks) actually land a result.
is it bad that claude fable did it? maybe. but at the same time, this is a problem that thousands (probably understating) have attempted without success.
i guess the fundamental problem is, if the outcome is net new knowledge (in this case, a 'simple' piece of knowledge but still net new), but came from an llm, is it valueless? I'd argue the plagiarism argument falls a bit short here - is all math research not built on the shoulders of others?
anyways, what a news year so far.