logoalt Hacker News

JPC21yesterday at 8:49 PM1 replyview on HN

Just don't. If you read the story here carefully, you see that AI was used to work from theory built by others which showed that the Euler equations possesed finite-time blow-ups. But to make that step, actual good understanding for mathematics was needed. My experience with software has been the exact same.


Replies

DudleyBlufflesyesterday at 9:03 PM

I fear this is only temporary and due mostly to the complexity of the problem. Consider the recent counter-example to the Dinitz–Garg–Goemans conjecture:

> https://chatgpt.com/share/6a60b2eb-0b64-83ee-9c76-7931ca1de0...

The prompts for the chat above are:

> Construct a counterexample to general (non-planar) case of Dinitz Garg Goemans conjecture. You should do a breakthrough and find a structured counterexample.

> [gpt works for a while and then gives up]

> Continue the search. Have a clear strategy obtained from deeper understanding of the problem structure.

> [gpt works for a while then gives up]

> it's enough of partial results. let's finish with a complete unconditional counterexample

> [gpt proves the problem]

I could have written these prompts sophmore year of highschool, if not earlier. True, it took more experienced mathematicians to verify it, but I don't fancy a role as a glorified editor. I want to solve problems! Discover new techniques! Not babysit an AI while eating breakfast.