Mathematicians want proof OpenAI didn’t use their work

Mathematicians are challenging OpenAI over the data used to train its AI models, particularly concerning its recent mathematical discoveries. Andreas Thom, a mathematician, suspects his interactions with ChatGPT may have contributed to OpenAI's announcement of a breakthrough in his field of non-sofic groups. He questions whether his unpublished work was used as training data, citing OpenAI's detailed command of specific, non-obvious techniques. Despite contacting OpenAI researchers, Thom found their response unsatisfying, as it only addressed direct access, not indirect use via de-identified training data. He argues that OpenAI, as the only entity with the relevant data, bears the responsibility to prove no misuse occurred. This incident, alongside a similar controversy involving another mathematician and OpenAI's recent Millennium Prize solution, has fueled concerns about transparency, ethical use of user data, and the potential for a secretive, competitive environment in mathematics.

(Source：The Verge)