posted in Technology

OpenAI's internal model solved over 700 longstanding open problems.

overview of all results is here:

github.com/openai/math/blob/main/overview.pdf

Several are fields medal worthy results. One proves the Riemann hypothesis for Re(z) >7/8, which is a substantial improvement over previous results

openai.com/index/sharing-ai-progress-in-mathematics/
enPage

Replying to @⁨FauxLiving@lemmy.world⁩

How many of those over 700 proofs have you evaluated yourself?

Whatever critcisms of the proofs by OpenAI is unlikely to be found in documents published by OpenAI.

Here’s one criticism of one proof:

scientificamerican.com/…/did-openai-solve-the-wro…

Edit: I’m not saying that none of those proofs are correct, but if you want a full and complete picture of what those proofs are and what they mean in context, you’re absolutely not going to get it from OpenAI. Mathematics is not “just trust me bro”, nor is it publishing press releases.

A wall of water falls along a cascading, seemingly human-made path.Scientific AmericanDid OpenAI solve the wrong Navier-Stokes problem?OpenAI’s proof seems eligible for a $1-million prize—but only by using a controversial loophole