Google Deepmind's AlphaProof Nexus solved dozens of open math problems and actually got peer reviewed

AI now reliably solves advanced math problems, moving beyond assistance to genuine contribution.

Google Deepmind's AlphaProof Nexus has achieved a significant breakthrough by solving numerous open mathematics problems, including nine previously unsolved Erdős problems and 44 conjectures from the OEIS database. Two of the solved Erdős problems had remained open for 56 years. The AI system, built upon Gemini 3.1 Pro, drafts and revises mathematical proofs using the formal language Lean, which provides automated error checking and guidance. For particularly challenging sub-problems, AlphaProof Nexus can leverage a specialized tool called AlphaProof. Notably, the system successfully tackled the nine Erdős problems even without resorting to AlphaProof. The research addresses ongoing debates in the AI and mathematics community, with Deepmind opting for peer review for its findings, contrasting with other recent AI-generated proofs released directly online.

The successful peer review of AlphaProof Nexus's mathematical proofs marks a critical juncture in the integration of AI into formal academic research. By submitting its work to rigorous scrutiny, Deepmind signals a potential path for AI-generated discoveries to gain broader acceptance within established scientific communities, addressing concerns raised by mathematicians about the validity and integrity of AI contributions. This contrasts with a trend of immediate public release of AI proofs, which has fueled skepticism and calls for boycotts.

This development suggests that AI systems are maturing from tools that assist human researchers to independent engines of discovery capable of pushing the boundaries of knowledge. The dual approach of using Gemini Pro for drafting and Lean for verification, augmented by specialized tools, highlights a robust architecture for complex problem-solving. The AI's ability to solve problems previously intractable for humans will likely accelerate research across various mathematical fields and potentially inspire new AI architectures for other scientific disciplines.