The mathematics community recently began voicing serious concerns regarding OpenAI's actions concerning authorship, data privacy, academic plagiarism, and transparency.
In August, OpenAI gathered about forty prominent mathematicians to discuss topics related to how artificial intelligence is beginning to surpass human capabilities in this field. During these meetings, OpenAI hinted that it had managed to solve hundreds of complex mathematical problems that had previously remained unsolved by humans.
However, by September, serious disagreements arose regarding how OpenAI claims results, how user data is processed, accusations of 'intercepting' others' work, and the general lack of process transparency.
A specific conflict emerged around the famous Millennium Prize Problem—namely, the existence and smoothness of the Navier-Stokes equations, the solution to which carries a one-million-dollar reward. Mathematicians Tristan Backmaster and Levent Alpoğlu used OpenAI tools to assist in a complex investigation of this problem.
Before they could publish their findings, OpenAI announced that its internal model had already solved the same aspects of the problem. On September 8, OpenAI released a statement saying: 'Since August 28, we have been training a new internal model that has demonstrated unprecedented performance in our benchmarks, including mathematics. Training this model continues, and its performance continues to improve.'
The company added: 'We are sharing the solution to the existence and smoothness of the Navier–Stokes problem, one of the Millennium Problems. This proof, created by the internal OpenAI system, shows that the dynamics of Navier-Stokes equations for fluid motion can develop a singularity in finite time. We provide both a description of the proof and its formalization in Lean.'
This meant that the painstaking work of Backmaster and Alpoğlu would no longer be recognized. A more serious issue was that they alleged OpenAI used their private chat sessions to accelerate its own internal research. Shortly thereafter, Backmaster published a statement reporting that OpenAI had given him a strict condition: either publish jointly with the company or allow OpenAI to publish the material first.