Measuring AI's ability to autonomously solve problems that arise naturally in the mathematical research process.
The First Proof Project provides independent assessments of the reasoning abilities of AI systems in the context of research mathematics. We present sets of research-level math questions — drawn from fields such as algebraic combinatorics, spectral graph theory, algebraic topology, and more — for AI systems to solve autonomously.
- 🌐 Website: 1stproof.org
- 📄 First Batch Paper: arXiv:2602.05192
- 💬 Community Discussion: Zulip channel (hosted by ICARM)
- 📧 Contact: contact@1stproof.org