First Proof: Research-Level Math for AI Evaluation (1stproof.org)

Researchers have announced the launch of the "First Proof" project, introducing a set of ten research-level math questions designed to assess the capability of AI systems in solving complex mathematical problems. These questions span various fields, including algebraic combinatorics, spectral graph theory, and more, drawing from real scenarios encountered during the authors' research. The project aims to establish a structured benchmark for AI problem-solving, inviting the AI community to engage with these questions and share results under the hashtag #1stProof before the encrypted solutions are revealed on February 13, 2026. This initiative is significant for the AI/ML community as it seeks to foster deeper interaction between AI systems and advanced mathematical concepts, potentially enhancing their problem-solving abilities in research contexts. By allowing AI practitioners to test these challenging questions, the project not only encourages collaboration and innovation but also sets the stage for developing more effective evaluation metrics in AI research. This could lead to breakthroughs in how AI systems understand and tackle complex problems traditionally reserved for advanced human mathematicians.