ഓപ്പൺ എഐയുടെ പുതിയ ഗണിത ഗവേഷണങ്ങൾക്കെതിരെ ഗണിതശാസ്ത്രജ്ഞർ രംഗത്തെത്തിയതോടെ, എഐ ഗണിതശാസ്ത്രത്തെ മാറ്റിസ്ഥാപിക്കില്ലെന്ന് ഗൂഗിൾ ഡീപ്മൈൻഡ് വൈസ് പ്രസിഡന്റ് പുഷ്മീത് കോഹ്ലി വ്യക്തമാക്കി. എഐ ഗവേഷകർക്ക് സഹായകമായ ഒരു ഉപകരണം മാത്രമാണെന്നും, ഓരോ ഉത്തരവും പുതിയ ചോദ്യങ്ങളിലേക്ക് നയിക്കുന്നതിനാൽ ഗണിതശാസ്ത്രം ഒരിക്കലും പൂർണ്ണമായി പരിഹരിക്കപ്പെടില്ലെന്നും അദ്ദേഹം പറഞ്ഞു.
After the Association for Human Mathematics called for a boycott of OpenAI over its latest maths research release, Google DeepMind vice-president Pushmeet Kohli says advancing mathematics is about empowering researchers with AI tools, rather than solving the field entirely. He says AI can help mathematicians find proofs and explore new questions, but mathematics will never be completely solved.
In a series of posts on X, Kohli outlined DeepMind’s approach to mathematical research, saying AI should help mathematicians advance the field rather than replace their work. “I have always believed that advancing mathematics is about empowering mathematicians,” he wrote. “From our earliest days at @GoogleDeepMind, our focus has been on deep collaboration with the mathematical community, pioneering new tools alongside one another to push the boundaries of what’s possible.”
His comments come after the Association for Human Mathematics called on mathematicians to stop working with OpenAI following the company’s release of findings on 377 mathematical problems and a repository containing more than 700 files. The group criticised OpenAI’s approach to AI-generated mathematical research, saying mathematicians had not asked for the work and raising concerns about the norms of scientific research.
Kohli argues that AI will not bring mathematics to an end, even as it becomes better at finding and checking proofs.
“To say that we are living through a transformational time for mathematics is not an overstatement. AI's growing ability to find proofs coupled with ability to check them quickly due to the pioneering work of the formal mathematics community is leading to remarkable answers. But mathematics will never be "solved": every answer opens up new questions,” he wrote.
Kohli added that this was just the beginning, with AI creating opportunities to ask more important questions and understand the answers.
Kohli also highlighted how DeepMind has worked with mathematicians to develop AI tools that can help them explore mathematical problems. He pointed to an early collaboration with researchers from the University of Sydney and the University of Oxford, which led to a 2021 paper in Nature on how AI can help mathematicians come up with new conjectures.
According to Kohli, AI can help mathematicians explore new ideas and solve complex problems rather than replace them. He said feedback from mathematicians also helped DeepMind develop AI tools such as AlphaEvolve and AlphaProof. Kohli also credited the formal mathematics community for developing tools such as Lean, which allow researchers to verify mathematical proofs.
Why mathematicians are calling for an OpenAI boycott
Meanwhile, the Association for Human Mathematics has criticised OpenAI’s latest maths research release, which includes findings on 377 mathematical problems and more than 700 files. The group described the release as “not a demonstration of scholarship, but a demonstration of power”. It said mathematicians “did not ask for this work to be done” and accused the company of disregarding established norms of scientific research.
An independent advisory group on mathematics and AI, hosted by the Institute for Advanced Study in Princeton, has also raised concerns about AI-generated mathematical arguments. It warned that researchers may struggle to understand or verify these arguments, or take responsibility for them. The group also cautioned that unequal access to advanced AI models and computing resources could give researchers at AI companies an advantage over others.
OpenAI, however, has said it consulted the advisory group and followed its advice and public recommendations when releasing its work. The company has also shared its findings in a GitHub repository, including formal versions of many proofs in Lean and summaries of the model’s reasoning.