IT Home October 10 news — According to a report by Business Insider this morning (the 10th), the large batch of mathematical research results recently released by OpenAI has drawn mixed reactions in academia. Some scholars are excited about the breakthroughs in mathematical research, while others have begun to worry about researchers who are still working hard to solve the same problems.
Tristan Buckmaster, a mathematics professor at New York University, said in an interview on CNBC's Squawk Box program that the results released by OpenAI have severely damaged the research work of some young mathematicians: "They released this much all at once, and our entire research program has been destroyed。”
Buckmaster also raised another concern: researchers had submitted unpublished research materials to AI tools.Did this content help OpenAI achieve the mathematical results it has now published? "You have to understand, they have access to all of our grant proposals. After a proposal is submitted, review experts need to read it and write summaries. And when writing summaries, they often input the proposal content into models."
On the 6th local time, OpenAI published more than 700 mathematical research manuscripts on GitHub. The announcement stated that these results came from a frontier AI model developed within the company. Many mathematical proofs also come with computer-verifiable versions.
For some mathematicians, the problem is not only thatthe problems they labored over were solved first by others。
Bryna Kra, a mathematics professor at Northwestern University, pointed out that AI helps advance mathematical research, but releasing a large volume of results all at oncecould undermine the long-standing tradition of collaboration in the mathematical community, and this collaboration is precisely an important foundation for research progress. "Mathematics is an extremely collaboration-oriented discipline. No one at OpenAI is capable of giving an academic talk on any one of these results, or answering technical questions."
Mathematicians often share unfinished research ideas in academic talks and exchanges. Kra worries that if people fear their ideas will be absorbed by AI before they can be formally published,they may no longer be willing to communicate openly in the future。
According to IT Home's understanding, OpenAI's announcement stated that norms for paper revisions and citations have been established for the mathematical results being released. The company also plans to improve the presentation of the papers and to fund seminars to help researchers understand the AI-generated mathematical results.
Alex Kontorovich, a Distinguished Professor of Mathematics at Rutgers University, holds a more optimistic view: AI is prompting academia to reconsider how to recruit talent and award prizes, and the value of mathematical training may actually increase further. "I think,we won't want to give a prize to someone who can only press a button. The most important thing in learning mathematics is developing the ability to think clearly, deeply, and for long periods when facing major problems. I believe the demand for these abilities will be even greater in the future, because they are hard to master and require long-term training."