Mathematicians put AI model AlphaProof to the test

Ringer, Talia

2 min read Original article ↗
  • NEWS AND VIEWS

Mathematicians use computational tools to prove theorems. An AI model that is trained to use these tools might accelerate mathematical discovery.

By

  1. Talia Ringer

    1. Talia Ringer is in the Siebel School of Computing and Data Science, University of Illinois at Urbana–Champaign, Urbana, Illinois, 61801, USA.

Last year, Google DeepMind made headlines for building an artificial-intelligence system that achieved a silver-medal standard on the problems from that year’s International Mathematics Olympiad (IMO), a prestigious competition for young mathematicians held annually. It was the first time such a system had performed at a medal-worthy level in solving IMO challenges. Now, writing in Nature, Hubert et al.1 report the details of the core technology behind that achievement: a tool called AlphaProof, which is designed to prove mathematical statements.

Nature 651, 595-597 (2026)

doi: https://doi.org/10.1038/d41586-025-03585-5

References

  1. Hubert, T. et al. Nature 651, 607–613 (2025).

    Article  Google Scholar 

  2. Castelvecchi, D. Nature 644, 20 (2025).

    Article  PubMed  Google Scholar 

  3. Wu, Y. et al. Preprint at arXiv https://doi.org/10.48550/arXiv.2205.12615 (2022).

  4. Ringer, T., Palmskog, K., Sergey, I., Gligoric, M. & Tatlock, Z. Found. Trends Program. Lang. 5, 102–281 (2019).

    Article  Google Scholar 

  5. Kontorovich, A. Preprint at arXiv https://doi.org/10.48550/arXiv.2510.15924 (2025).

Download references

Competing Interests

The author declares no competing interests.

Subjects

Latest on: