Terence Tao: identifying a promising problem is now the scarce resource, and 'even the rumor of someone working on a problem' triggers AI effort to flatten it
In a four-post Mathstodon thread on 2026-09-08 (193 favourites, 123 boosts), Tao argues open math problems are being mined non-renewably: problems are infinite, but fruitful ones are not, the way a country can lack drinking water while surrounded by ocean. His specific mechanism is that every new tool flattens a field's difficulty landscape, and the AI era is unusual in having no visible frontier separating AI-feasible from AI-hard problems, worsened by labs not disclosing negative results or their process. His conclusion is a governance one: the incentive is now to stop sharing research directions publicly, and he proposes designating classes of problems where a raw solution without analysis has negligible or negative value.
↳ Follow the thread