Which Millennium Prize Problem will AI solve next?

The strongest reason to keep this outcome above a token probability is that BSD sits in a part of mathematics where AI-assisted progress can plausibly accumulate into a breakthrough. The conjecture is deeply structural, but it is also connected to elliptic curves, ranks, L-functions, and computational experimentation, which are all areas where modern models and automated theorem-proving tools can contribute meaningfully. Recent chatter about AI labs, especially around record-setting elliptic curve computations and broader math research efforts, suggests that BSD is viewed inside the market as one of the more plausible Millennium problems for a lab to claim first.

At the same time, the evidence available today is still far from an actual solution. The public record does not show a verified proof, no official lab announcement exists, and there is no clear sign of a near-term consensus theorem that would make a formal solution announcement likely before the end of 2027. BSD is also notoriously hard because its central claim links algebraic and analytic objects in a way that has resisted decades of expert work. Even if AI finds important lemmas or counterexamples to proposed criteria, turning that into a full qualifying solution is a much higher bar than producing impressive computations or partial results.

Market pricing at around 15.5% for Yes suggests investors think BSD is one of the more plausible candidates among the five problems, but the current story is still mostly rumor, inference, and extrapolation from AI progress rather than evidence of an imminent proof. That makes the market’s Yes probability understandable, yet still somewhat optimistic. My view is that the base rate for any one specific Millennium problem being solved by an eligible AI lab before 2028 remains modest, and BSD’s combination of deep theory, verification burden, and limited public signals keeps it below the market price, though well above single-digit territory.

The main upside path is a credible lab announcement that presents a complete proof and is accepted as a qualifying solution without needing formal Clay endorsement. The main downside path is that AI continues to generate useful partial results, but none cross the threshold into an official solution, leaving BSD unresolved by the cutoff date. Because the market only cares about the first qualifying AI-lab announcement, even a genuine breakthrough on another problem would make BSD miss this contract, which further reduces the practical odds.