DeepMind's Irving Assigns High Probability to Unbounded-Depth CSETH

DeepMind researcher Geoffrey Irving raised his probability estimate for the unbounded-depth version of CSETH, a formal math benchmark conjecture, to over 90%, after previously giving it a greater-than-80% probability despite losing Bayesian points on related ETH questions.

2026-10-28 ~ 2026-10-29 · 2 related posts

• Geoffrey Irving puts >80% on CSETH, signaling rapid AI formal-math progress — geoffreyirving · 2026-10-28
• Geoffrey Irving puts >90% on unbounded-depth CSETH, unsure about finite depth — geoffreyirving · 2026-10-29