AI outperforms humans in all mathematical research areas by 2028?

Prediction market on manifold. Will I believe, 3 years from today, that publicly available AI tools have absolute advantage over human mathematicians in all areas of mathematical research? I will not participate in this market. Some background--I'm a math professor at the University of Toronto, mostly focusing on algebraic geometry and number theory. I will resolve this market YES if I think publicly available AI tools outperform most professional mathematicians (e.g. me) as of July 17, 2028, and NO otherwise. In particular, for a YES resolution, they should be able to create high-quality original research with very little human input. See here for a related market, with a more precise operationalization of a related question, albeit with some minor problems as a proxy: https://manifold.markets/TamayBesiroglu/will-ai-be-capable-of-producing-ann Update 2025-07-17 (PST) (AI summary of creator comment): The creator has specified that for a YES resolution, the AI must outperform humans in all areas of math research, which explicitly includes: Creating definitions Making conjectures Update 2025-08-09 (PST) (AI summary of creator comment): - If the creator is unreachable at resolution time, the market should resolve to N/A.

Liquidity: $1,600. Resolves: 7/18/2028.