Terence Tao Responds to the OpenAI Math Drop

Commenters reacted to Terence Tao’s critique of AI “drop-in” proofs by debating what value remains in mathematics when large language models can produce solutions. Several worried, as shubhamjain and AlexAplin put it, that simply dumping machine-produced proofs “contaminates” the search for alternative, insightful approaches and rewards being first over community building and exposition. Others urged protocols for using AI in STEM (Regex777) or pointed to historical analogies - Ramanujan (koopuluri) and Thurston (sanxiyn) - suggesting the social work of explaining and integrating results could fall to humans. Some saw opportunity: contubernio argued that raw technical grunt work will give way to premium on broad perspective and creativity, while others like bluepeter and MrOrelliOReilly think future models will also handle exposition and exploration.

Opinion divides on consequences. A number of commenters warned of disrupted careers and doctoral training (doctoboggan, cs_throwaway) and growing inequality because AI resources are costly (contubernio). Others worried math could accelerate beyond human understanding, creating machine-frontier and human-frontier split (OtherShrezzing). Parallels to software were drawn (lifeisloving, lubujackson), with some saying AI will make routine creation boring unless we change how we integrate tools. A few suggested concrete next steps, from formalizing what counts as “interesting” problems (adrianN) to new collaborative, structured research models.