> At 11:58 P.M. EDT on Monday, Buckmaster posted his and Alpöge’s results, along with the statement laying out his account of events. According to the statement, he does not know how the OpenAI team so quickly reconstructed the details of its approach. Buckmaster wrote that he asked Bubeck if the OpenAI LLM was given access to his personal prompts to the company’s model Codex, which he and Alpöge were using throughout the process and thought were private. Bubeck told him no but wouldn’t answer whether the model’s training involved user data, which might have included Buckmaster and Alpöge’s private prompts, Buckmaster alleged.

Second to the achievement, I think this is the big takeaway and should be approached is if they did, in fact, use private prompts for training.

Mathematicians vs big business. It sounds ugly.

Navier-Stokes blow-up counter example found.

They stole it. No chance this was coincidental. The path is pretty unique and Buckmaster has a provable trail to support his story.

The worse thing about this fiasco, is that we now not know if the previous breakthroughs attributed to "AI" also involved stealing some user's partial work towards them.

Academia is all about lying and stealing, but this is just too blatant, even for their standards.

> they’ve been pushing their models to do math no human can, faster than any human can. It appears they’ve succeeded.

Fuck you they have.