Post

It's surely just regurgitating from the many counterexamples to the Jacobian Conjecture that were in the training data

Everyone is just taking that tweet at face value? Assuming that result is correct, we don't currently have verification it was done just recently - vs. a project they secretly devoted months to then chose to reveal at the most viral moment. Meanwhile, plenty of outside proof for both positions.

sorry but "they actually had this outcome a while ago and sat on it" is a really silly conspiracy theory. I cannot think of an upside and your links are totally unrelated? It's not a surprise that LLMs are bad at news, especially without context? imo very much a misuse of the technology

I don't think this is what happened here, but I could imagine a company throwing a lot of compute at math problems, sifting out the good results, and then drip feeding them out to keep enthusiasm up and balance bad PR. Seems like a pretty sensible strategy to me, actually.