AI is dividing the math world in two

AI labs are racing through academic open problems so fast that peer review, attribution, and researcher trust are breaking down behind them.

• OpenAI's math blitz: OpenAI spun up roughly 10,000 compute agents at a rumored cost in the tens of millions to solve a Millennium Prize problem in about 88 hours after hearing rival researchers were close.
• Mass, messy drop: OpenAI released nearly 400 results across 700+ preprints in one evening; three papers were already retracted for a sign error, and mathematicians say full review could take years.

Themes, sentiment and reasons are AI-written paraphrases, checked so they never quote an episode. They may contain errors. Not investment advice. Episodes remain the property of their creators.