seeing a very interesting LLM psychosis case brewing on Twitter, this one guy David Budden (a well-respected programmer, former Director of Engineering at DeepMind) is convinced he can solve Millenium problems with bounties just by using LLMs and feeding DAGs into them (seemingly), and experienced mathematicians are trying to loot him by tricking him into 5-figure bets (based).

he's released a bunch of proof papers so far, and he's lashing out at anyone speaking against his proofs and demanding they provide a long clear explanation of why his results aren't valid (most of it is him declaring new axioms), even as he admits he is not a math guy.

is LLM psychosis a real thing or do LLMs really appeal to people already having a tenuous grasp on reality