By Dave Kung

AI is everywhere. Or at least talk of AI is everywhere around the mathematics community. Frontier models are radically reshaping research mathematics, with over 700 papers (some groundbreaking) released on a single day and the Navier-Stokes problem solved (with a disputed amount of human assistance ). Twenty-four Fields medalists collaborated on a statement about the “ severe misalignment of AI and mathematics ” while one of this year’s recipients used the press conference to announce his pivot to AI safety 1 .

We face a similarly unsettling situation when it comes to the impact of AI on mathematics education. Will students use AI to super-charge their learning, using new tech tools to get individualized help, micro-targeting their particular struggles and confusions? Or will they use it to short-circuit their learning, saving time in the short-term at the expense of longer term understanding?

A few months ago I was legitimately unsure.…