# AI is Revolutionizing Mathematics AI is disproving mathematical proofs and discovering new theorems. - **Source:** [Hacker News](https://news.ycombinator.com) - **Published:** 2026-07-20 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/68e28a15-83cb-4e61-8db6-9fb04d8a348b ## A New Era in Mathematics AI tools have begun disproving mathematical conjectures and discovering new theorems. ChatGPT debunked Erdős's Unit Distance conjecture, marking a first in this field. This development has caused significant waves in the mathematical community. ## Formalization with Lean The proof generated by ChatGPT was formalized using Lean, an interactive theorem prover. Logical Intelligence automated this process, demonstrating how AI can handle mathematical arguments. This is a major step forward for future mathematical research. ## Global Class Field Theory Disproving Erdős's conjecture required formalizing complex mathematical structures like global class field theory. AI processed this 100+ page proof, showcasing its ability in areas where even mathematicians struggle. ## OpenAI's Role OpenAI's new model, Sol, automatically formalized the disproof of Erdős's conjecture. Led by Boris Alexeev, the team generated 1.2 million lines of Lean code, highlighting AI's potential in mathematical research. ## Reliability of AI-Generated Code The reliability of AI-generated code was questioned. Mathematicians tested the code in sandbox environments to verify its accuracy. Lean being a programming language also raised concerns about potential malicious code. ## Fermat's Last Theorem AI attempted to formalize the theory of finite flat group schemes needed for Fermat's Last Theorem. However, an error was found in an LLM-generated document, showing AI still requires human oversight. ## Logos Research's Contribution Logos Research developed a tool to translate mathematical texts into Lean. This tool supported mathematicians working on Fermat's Last Theorem, though access was limited, leading participants to alternative AI models. ## Claude Max and ChatGPT Pro Participants at the Formalizing Fermat workshop had access to AI models like Claude Max and ChatGPT Pro. These models were optimized for mathematical research, with OpenAI's Sol model becoming available towards the end. ## AI and Mathematical Discoveries AI is increasingly playing a role in mathematical discoveries. Models like ChatGPT can find counterexamples that human mathematicians might miss, potentially changing the nature of mathematical research. ## Mathematicians' Reactions Many mathematicians are skeptical about the reliability of AI-generated proofs. However, figures like Fields Medalist Mike Freedman acknowledge the technology's potential. The mathematical world is adapting to this new era. ## Future Developments AI-assisted mathematical research will accelerate in the coming years. Tools like Lean and more powerful models will drive progress. Mathematics is entering a new phase with AI. ## Conclusions and Predictions AI has the potential to revolutionize mathematical research. Tools like ChatGPT and Lean can disprove complex theorems and make new discoveries. However, more work is needed on the reliability and ethical implications of this technology. ## Sources covering this story - https://xenaproject.wordpress.com/2026/07/20/human-mathematicians-are-being-outcounterexampled/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._