If I'm not mistaken, a lot of math researchers work at openAI full time on using LLM's to tackle open research problems