On October 9, it was reported that OpenAI released over 700 preprints on GitHub on October 6, describing significant mathematical achievements made by its internal AI models. Less than 24 hours later, the company retracted three of the manuscripts due to an error. OpenAI published 722 preprints, claiming progress on 372 mathematical problems across geometry, computer science, algebra, and other fields. Hundreds of AI-generated mathematical discoveries advanced the frontiers of higher mathematics in just one day. However, on October 7, OpenAI announced the retraction of three manuscripts due to a symbol error. Mathematicians have expressed dissatisfaction and concern over OpenAI's actions thus far. The Human Mathematical Society also criticized the recent release, stating on October 7, 'Mathematicians did not request this work. The release of over 700 documents at once is not an academic showcase, but a display of power. We urge mathematicians to cease collaboration with OpenAI and return to a science vision centered on human understanding.' Fields Medalist Terence Tao shared the statement.
All Comments