Kenneth Stanley

@kennethstanley.bsky.social

SVP of Open-Endedness at Lila Sciences. In the past: Maven CEO, Lead at OpenAI, head of basic/core research at Uber AI, professor at UCF. Stuff I helped invent: NEAT, CPPNs, HyperNEAT, novelty search, POET, Picbreeder. Book: Why Greatness Cannot Be Plann

New AGI benchmark proposal: instead of giving the AI a hard test like AIME, have the AI come up with a corresponding hard test of its own (and answer key), and then give that test to human students of the right level. Was it a good test with novel questions?