We're gonna need a lot more mathematicians

Amit Sahai argues that rapid advances in AI will soon generate a flood of novel, consequential mathematical and scientific ideas that far outpace individual human researchers. Rather than retreating when machines produce insights faster than humans can absorb them, society must invest in collective human understanding so decisions of enormous consequence remain interpretable and contestable. He gives a concrete risk: an AI could propose a radically new terawatt fusion-plant design whose safety depends on unfamiliar models and failure modes; approving construction should require broad human comprehension of the model, its empirical support, and its uncertainties, not blind trust in opaque machine reasoning.

To meet this need, Sahai calls for a major expansion of mathematically sophisticated people organized into sustained, well-supported teams that jointly study and validate AI-generated breakthroughs, using AI as a tool to help, not replace, human understanding. He labels this a “deployable intellectual reserve” and urges mathematicians to broaden their vocation beyond narrow research topics to include public-facing technical appraisal. He rejects the idea that AI alone will allow a tiny number of humans to suffice, noting human limits of time and depth; collective, distributed effort will be essential if humanity is to retain meaningful agency over AI-driven innovation.