@tonylfeng
Math professor at UC Berkeley, Research Scientist at Google DeepMind
AI
A poster argues that AI has made publication less reliable as a stamp of correctness or prestige. They propose judging papers regardless of provenance and requiring video talks to show human understanding.
AI
A workshop participant says the report follows a two-day Simons Institute discussion on how the theoretical computer science community should adapt to rapid advances in AI.
AI
A post says Carlo and Mark’s paper establishes consistency properties for probabilistic models of number-theoretic invariants. Its author credits a model their team post-trained in early July with solving the problems the paper addresses.
AI
A post introducing the work claims 4.3× higher “interestingness,” with a quantitative notion of that quality and a self-expanding discovery loop.
AI
The host’s September 9 announcement said the discussion would examine what would count as a verified solution and what such a result could mean for artificial general intelligence.
AI
One post connects the question of knowledge’s usefulness to the step-by-step automation of knowledge production.
AI
Posts on X highlight a claimed solution to a 1948 topology problem using Claude.
