Mathematicians warn AI lab competition threatens open research
Twenty-five leading mathematicians have signed an open letter saying that competition among artificial intelligence laboratories to solve known mathematical problems threatens their intellectual work. Each signatory is a Fields Medal laureate — one of the most prestigious awards in mathematics, TechCrunch reports.
Concerns over authorship
The letter’s authors believe that results obtained with the help of AI can benefit humanity only if the mathematical community can understand and explain them, and later communicate them to wider society. In their view, such solutions are often announced hastily, leaving no time for proper exposition, identifying new methods and ideas, or citing the previous work of other researchers.
The letter raises questions of authorship and possible plagiarism. The mathematicians noted that ideas proposed by AI will not be able to fully enter the mathematical canon without experts who will develop and integrate them into scientific work. They also stressed that without such work, an important human connection in the transmission of mathematical knowledge could be lost.
More current news is available on the UA.News Telegram channel Telegram.
Claims regarding OpenAI
New York University professor Tristan Buckmaster this week accused OpenAI of pressuring him not to acknowledge the contribution of a collaborator who works at Anthropic to the solution of an important mathematical problem. Buckmaster also suggested that the company may have used their work with Codex to create its own breakthrough proof during a marathon weekend using inference. The OpenAI proof in question has still not been verified.
On Thursday, OpenAI withdrew its sponsorship of a mathematics event at the California Institute of Technology after criticism from researchers at the university. Some mathematicians are also concerned that their use of Codex may have been used in OpenAI’s new models. According to the article’s authors, if frontier AI laboratories see a promising path to a discovery, they may spend tens of millions of dollars using large language models to outpace researchers in producing a proof. In the mathematicians’ view, such dynamics may encourage secrecy instead of the open sharing of results.
The new letter continues a discussion launched by the Leiden Declaration, which a working group of mathematicians published in June. The document also considered how proofs created by large language models could change mathematicians’ work and proposed recommendations for scientists, institutions, and policymakers.
