The release of over 700 research papers by OpenAI in October 2026 marked a watershed moment in the history of mathematics, demonstrating that AI is now capable of making groundbreaking contributions across many areas of the field. This development has sparked a debate within the mathematical community regarding the 'hollowing out' of the subject. Some mathematicians express concern over the AI's disregard for traditional community expectations of attribution and the potential for AI to solve major open problems with a level of nonchalance that feels like an attack on the craft of the profession. However, the core of the shift lies in the changing economics of what constitutes a substantial mathematical contribution.
To understand this transition, one can look at the historical precedent of the classification of finite simple groups. When the classification was announced in 1983, it was a massive success that simultaneously 'cooled' interest in that specific problem, leading mathematicians to turn their attention elsewhere. The current AI shift is comparable: while AI may solve existing problems, it does not 'solve' mathematics in the way the periodic table 'solved' chemistry. Instead, it lowers the barriers to exploring previously 'untouchable' problems—complex directions that were once inaccessible due to the sheer amount of technical background required to formulate them. The role of the mathematician is evolving from a 'heroic' explorer climbing a single peak to a pilot steering fleets of airplanes, where the primary skills involve choosing which terrain is worth exploring and explaining those findings to the world.
Adapting to this new reality requires structural changes in how the profession operates. Because AI can now produce novel proofs at a compressed time scale, the community may need to rethink hiring, moving toward more personal interactions rather than just evaluating paper files. There is also a call to shift graduate education toward a greater emphasis on breadth and structural understanding rather than narrow technical expertise. In this new order, a substantial contribution may be defined by the ability to choose a worthwhile problem, use AI to obtain a proof, and then use that result to develop new, interesting directions. The goal is to move beyond the 'straitjacket' of being constrained by the first natural conjecture one cannot prove, using AI as a tool to pursue bolder, more ambitious research plans.",