OpenAI takes action: a committee of mathematicians will assess ChatGPT’s findings
Following allegations of plagiarism and a letter of protest signed by 25 Fields Medal winners, Sam Altman’s company has set up an independent advisory group to verify the model’s claims and scientific findings.
A group of mathematicians has been asked to review the results produced by ChatGPT in this field. OpenAI made the announcement following the controversy of recent weeks with the scientific community.
One of the areas of contention between the giants of artificial intelligence concerns the ability of their respective models to solve complex mathematical problems. In the last two months alone, Anthropic announced that Claude had solved the Jacobian conjecture, a mathematical problem that had remained unsolved for almost ninety years, whilst OpenAI claimed that ChatGPT had solved the Navier–Stokes problem relating to the motion of fluids.
This announcement has drawn criticism from part of the scientific community towards the company led by Sam Altman. Meanwhile, as reported by Axios, Tristan Buckmaster, a mathematician at New York University, and Levent Alpöge, a researcher at Anthropic, were working on the same topic, also using Codex, OpenAI’s programming model. And Buckmaster, in particular, essentially accused the company of plagiarism. The company described this as ‘unlikely’, but at the same time admitted that it could not rule out the possibility that ‘anonymised data from their use of ChatGPT may have helped to improve our model’.
Not only that: no fewer than 25 winners of the Fields Medal – the equivalent of the Nobel Prize in mathematics – have signed a letter condemning this ‘high-handed’ use of artificial intelligence. They point out that the aim of mathematical research is to ‘understand the fundamental structures of forms, numbers and natural phenomena’. It is certainly not to decree which is the best AI model on the market.
In response to this appeal, OpenAI has announced that it has set up an ‘independent advisory group on mathematics’, tasked with providing ‘advice on the review and communication of new findings’. The group, currently comprising nine mathematicians from American and European universities, will operate independently; it will be able to offer advice even without a direct request from the company, ‘comment on OpenAI’s impact on mathematics and make its views public’. However, it will not be able to “advise us on the pace of our internal progress in the field of mathematics”. In short, the idea is that this group will be able to publicly scrutinise announcements regarding the results achieved by ChatGPT in the field of mathematics. Only time will tell whether this will be enough to allay the scientific community’s criticisms.
