EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Harness design matters for AI capabilities in mathematics

Mohammed Abouzaid, Nikhil Srivastava, Rachel Ward, Lauren Williams iD

Proceedings of the National Academy of Sciences · 24.11.2026 · 123 (47)

DOI: 10.1073/pnas.2627448123 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 20.11.2026. Verfasst von Mohammed Abouzaid und 3 weiteren Autor:innen. Beteiligte Einrichtung: Department of Mathematics, Stanford University, Stanford, CA 94305. Die Arbeit nennt 6 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: AI for Math Fund (Renaissance Philanthropy Fund), Survival and Flourishing Fund, Anthropic.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

To assess the ability of current AI systems to correctly solve research-level mathematics problems, we tested four AI systems on a set of ten problems in a broad range of mathematical fields; these problems arose naturally in the research process of the contributors. This document includes our methodology, and the results of our testing, including the assessment of human reviewers and a report of costs. We provide links to supplementary documents including the human solutions, the AI-generated solutions, and the referee reports and logs for the AI-generated solutions.

Schlagwörter

Stichwörter der Autor:innen

AIFirst Proofbenchmarkmathematics

Autor:innen und Institutionen

  • Department of Mathematics, Stanford University, Stanford, CA 94305.
  • Department of Mathematics, University of California, Berkeley, CA 94720.
  • Department of Mathematics, University of Texas, Austin, TX 78712.
  • Department of Mathematics, Harvard University, Cambridge, MA 02138.

ORCID-Profile: Lauren Williams

Förderung

  • AI for Math Fund (Renaissance Philanthropy Fund)
  • Survival and Flourishing Fund
  • Anthropic
  • OpenAI

Zitieren

APA

Abouzaid, M., Srivastava, N., Ward, R., & Williams, L. (2026). Harness design matters for AI capabilities in mathematics. Proceedings of the National Academy of Sciences, 123(47). https://doi.org/10.1073/pnas.2627448123

Vancouver

Abouzaid M, Srivastava N, Ward R, Williams L. Harness design matters for AI capabilities in mathematics. Proceedings of the National Academy of Sciences. 2026;123(47). doi:10.1073/pnas.2627448123

BibTeX
@article{Abouzaid2026,
  author = {Abouzaid, Mohammed and Srivastava, Nikhil and Ward, Rachel and Williams, Lauren},
  title = {Harness design matters for AI capabilities in mathematics},
  journal = {Proceedings of the National Academy of Sciences},
  year = {2026},
  volume = {123},
  number = {47},
  doi = {10.1073/pnas.2627448123}
}
Kostenlos starten →