Which Millennium Prize Problem will AI solve next?
BSD remains a real but not leading candidate for the next AI-announced Millennium Problem solution. The market’s Yes price looks somewhat optimistic given that public signals point more strongly to other problems, especially Hodge, rather than BSD.
Analysis
The Birch and Swinnerton-Dyer Conjecture is still officially unsolved, and there is no public evidence that an eligible AI lab has announced a full solution. The recent context mainly shows adjacent progress: AI has been used in elliptic-curve work, and there is speculation that BSD could eventually benefit from such tools, but speculation is not the same as a qualifying solution announcement. For this market, the key hurdle is not just mathematical progress but a very specific public declaration by one of the listed eligible labs before the deadline, which is a much narrower and rarer event than general research advancement.
Arguments for Yes center on the fact that BSD is a natural target for AI-assisted mathematics because it connects deep number theory, elliptic curves, and computational experimentation. If an AI system helps discover a proof strategy, a counterexample to a key criterion, or a proof outline that the lab is willing to publicly claim as a full solution, BSD could qualify quickly once the right insight appears. The broader environment is also favorable in the sense that frontier AI labs are increasingly willing to announce major mathematical breakthroughs, so a breakthrough in an unsolved Millennium problem is no longer purely speculative science fiction.
Arguments against Yes are stronger at the moment. Public discussion has tended to center more on Hodge than BSD when “an unnamed Millennium problem” is referenced, so BSD does not appear to be the front-runner in visible reporting. BSD is also a notoriously delicate conjecture whose full proof would likely require deep new ideas rather than incremental AI-assisted computation alone, and many AI math results to date have been important but not of the final, problem-solving kind. The event definition is strict: the announcement must clearly claim a solution, come from an eligible lab or representative, and stand on its own without later confirmation, which reduces the chance that a near-miss or partial advance will count.
Overall, the market’s Yes price of 17.5% seems a bit high relative to the public evidence. BSD is plausible enough to justify a nontrivial probability, but it is not currently the best-supported next AI-solved Millennium problem, and the combination of proof difficulty, announcement strictness, and competing candidates makes a low-to-mid teens probability more defensible.
Arguments
For
- Arguments for Yes: BSD is closely connected to active AI-assisted work in elliptic curves and arithmetic geometry.
- Arguments for Yes: A single decisive proof announcement from an eligible lab would be enough to resolve the market.
Against
- Arguments against Yes: Public reporting currently points more strongly to other Millennium problems than to BSD.
- Arguments against Yes: BSD remains a very deep conjecture, and no eligible lab has publicly claimed a solution.
Key drivers
- AI labs are increasingly capable of making credible high-level mathematical claims.
- BSD has strong relevance to areas where AI-assisted computational insight may help.
- Public reporting currently favors other Millennium problems as more likely near-term targets.
- A valid event requires a full public solution announcement from an eligible lab, not just progress.
Risk factors
- A surprise breakthrough in elliptic curves or number theory could quickly elevate BSD.
- A lab may announce BSD as solved before broader verification if it believes the proof is complete.
- Public speculation may be underestimating how quickly frontier labs can move on open problems.
- The market could reprice sharply if another AI math breakthrough builds momentum toward BSD.
Scenarios
Best case
An eligible AI lab discovers a complete proof of BSD, publicly announces it as a solved problem, and the announcement is widely treated as qualifying before the deadline.
Most likely
AI labs continue producing meaningful progress and occasional speculation around BSD, but no qualifying public announcement for BSD occurs before the market closes.
Worst case
No eligible AI lab announces a qualifying solution to BSD by the deadline, and the next AI-solved Millennium problem turns out to be another problem or none at all.
More from this day
- politicsPolymarketEnded
Trump x Greenland deal signed by...?
AI12%MKT97%Edge-85HypedThe agreement appears to have been announced, but the formal signature is still described as coming next week rather than before the September 23 cutoff. That makes a Yes outcome possible only if the governments accelerate the signing unexpectedly, so I think the market is materially overpricing Yes.
- FinancialsKalshi13y
Will OpenAI or Anthropic IPO first?
AI23%MKT90%Edge-67HypedI think Anthropic is more likely to IPO before OpenAI, so the Yes side looks overpriced if Yes means OpenAI goes first. The clearest public signals point to Anthropic being closer to market, while OpenAI is explicitly signaling a later timetable.
- PoliticsKalshi2y
Who will Trump pardon?
AI3%MKT50%Edge-47HypedI think the chance Barron Trump receives a presidential pardon before Jan. 21, 2029 is very low, because there is no visible underlying federal offense or legal exposure that would make a pardon relevant. Absent an extraordinary future scandal, this looks far more like a speculative family-name market than a grounded clemency case.