In theory, math is pretty simple. You start with a bunch of axioms and logic rules and use them to prove increasingly complex statements. How hard could it be?
In practice, everybody knows that this simple description of math is basically useless. Doing any sort of interesting math requires creativity and intelligence. Right?
In this post, I want to posit that the recent AI improvements in math show that humans are just bad at math, just like AI has shown in the past decades that humans are just bad at games like chess.
Like math, chess is pretty simple in theory. And like math, everybody knows that in practice it's incredibly complex and requires creativity and intelligence to master. But since at least 1997, when Deep Blue beat Kasparov, it has become clear that humans are just bad at chess.
In what way are we bad at chess? Apart from the fact that no human is remotely capable of even getting a draw against a chess engine, we can't even comprehend their moves. The best way to see this is by watching the broadcast of a top chess event. Commentators (who are top players themselves) will often say things like "there's no way Player X finds this move, it's too unnatural" about the best engine move. And then Player X thinks for an hour and indeed plays a worse move. Think about that. Humans are so bad at chess that good moves are deemed "unnatural".
Now to get back to math, people are starting to say similar things about AI generated proofs. For example, "The "aha" insight for this is actually f**ing wild" on the proof of Barnette's Conjecture or It feels like something written by someone who’s on psychedelics on the proof of UGC. These proofs look sloppy and unnatural to experts, and some even doubt their correctness. But maybe this is just the first step in realising that humans are not good enough to keep up with the LLMs. I was not there in the early 2000s, but I can imagine chess player still doubting engine moves back then. They don't dare to anymore.
Time will tell if math, like chess and other games, is something that can be brute-forced with raw computation. And that the "creativity and intelligence" we have always thought was required is just our biological way of getting good at something that humans are inherently bad at. If it's the case, then maybe mathematicians will find solace in the fact that chess books are still being written and top players work with engines to get a better understanding of the game. Sure, the competitive aspect of chess is a big part in its enduring popularity, but a lot of people just find the game beautiful and want to understand it better.
Maybe we need to prepare for a world where it's clear that humans are bad at math, science, art, etc. Everyone needs to think about what they'll do in this world. I like to think I'll still enjoy math and chess.