A new preprint by researchers Wang and Wu from Hunan University presents a proof of the Spherical Hadwiger Conjecture, a problem that has remained open since the 1970s. Notably, the authors acknowledge using OpenAI's Codex to assist with developing proof details, identifying gaps, organizing the manuscript, and editing the English. While the researchers took full responsibility for the content and final decisions, the involvement of AI in such a complex mathematical proof represents a significant milestone.
The Spherical Hadwiger Conjecture is a specialized topic in integral geometry that has challenged mathematicians for decades. The recent proof, though not yet fully verified by the wider community, has passed initial scrutiny and is considered polished and readable by experts in the field.
This development underscores a broader trend: large language models and AI tools are increasingly capable of performing tasks traditionally reserved for highly trained mathematicians. OpenAI's earlier reports suggested that mathematicians' work is among the most exposed to disruption by AI, potentially affecting all aspects of their labor.
The implications extend beyond this single proof. As AI tools become more proficient, the role of mathematicians may shift from discovering new results to interpreting, communicating, and contextualizing them. This parallels how classical music has been institutionalized within conservatories, preserved as a cultural endeavor even as technology changes how music is produced and consumed.
Currently, pure mathematics largely exists within academic institutions, with limited career opportunities outside them. The question arises whether society will continue to support mathematicians in a similar way it supports classical musicians—valuing the preservation and communication of mathematical knowledge even as AI handles much of the technical work.
Ultimately, the integration of AI into mathematical research challenges existing incentive structures and raises important questions about the future of intellectual labor. Ensuring that human mathematicians remain relevant and supported may require rethinking how their contributions are valued in an era where computational intelligence can outperform human reasoning in many respects.