AI and Mathematics Should Align With Understanding, Not Institutions

Summary

Lior Pachter responds to the letter “A Severe Misalignment of AI in Mathematics,” signed initially by 25 Fields Medalists after OpenAI announced a solution to the Navier–Stokes existence and smoothness problem. The letter argues that AI companies prioritize rapid announcements over conceptual understanding, careful exposition, attribution, and the separation of genuinely new methods. Pachter agrees with that criticism but argues that the existing mathematics community is also poorly aligned with its stated goal of understanding because it has often failed to nurture students and ideas. He supports the point with historical and personal examples: Juliusz Schauder faced antisemitism, exclusion, wartime persecution, and murder; Olga Ladyzhenskaya was passed over for the 1958 Fields Medal despite foundational work in fluid mechanics; Karen Uhlenbeck encountered explicit sexism in academic hiring; and Cathleen Morawetz was met with a dismissive response when raising the underrepresentation of women. He also describes mentorship and discrimination problems at UC Berkeley, including the treatment of Julia Robinson and allegations concerning Yuval Peres, as well as demeaning comments directed at students and Yitang Zhang. The article questions whether Fields Medal status should confer special authority over mathematical responsibility, noting that prizes reflect judgments about which fields and styles of work matter. Pachter concludes that AI companies and mathematicians should align with understanding, attribution, intellectual generosity, and the careful nurturing of students and ideas, rather than simply reproducing existing incentives and hierarchies.