News Brief
OpenAI announced on September 21 that it has formed a mathematics advisory group, as its AI systems have solved more than 100 open mathematical problems. The advisory group will provide advice on research directions, but will not have the authority to slow down or change existing research processes. The move is intended to balance the breakthrough in AI mathematical capabilities with academic ethics, and may influence the future direction of AI applications in fundamental science.
Background
With the rapid evolution of generative AI’s reasoning capabilities, using AI to tackle mathematical problems has become a frontier direction in the industry. OpenAI’s formation of the advisory group is both a response to its AI achievements and a way to avoid the reputational risk of ‘working behind closed doors.’ Previously, AI had made progress in several projects in areas such as theorem proving and symbolic reasoning, but solving open problems on such a scale is a first. The mathematical community’s attitude toward AI participation in research has shifted from skepticism to cooperation, and the establishment of the advisory group may establish a new collaborative paradigm.
Deep Dive
By forming the mathematics advisory group this time, OpenAI appears on the surface to be ‘gatekeeping,’ but in reality it looks more like seeking endorsement for AI’s math report card. When AlphaGo defeated Lee Sedol, humans could still console themselves by saying ‘Go is not everything.’ Now that AI has solved more than a hundred open problems in one go, even the authenticity of the ‘solutions’ has to be confirmed by the advisory group—this in itself says something. Liu Gong believes that what really deserves attention is not how many suggestions the advisory group can offer—it cannot even pause research—but rather OpenAI’s choice to define the role of experts as ‘advisers’ rather than ‘reviewers.’ The next step is to see whether the mathematical community is willing to publicly verify these results, and whether AI’s ‘problem-solving’ in pure mathematics can be turned into real productivity in physics and computer science. If the answers to both are yes, then the division of labor among human mathematicians will likely need to be rewritten ahead of schedule.
Perspectives
Further Reflection
- Have the mathematical problems that AI claims to have solved undergone rigorous verification, and how genuine are they?
- Does the advisory group’s lack of authority to intervene in research processes reduce it to a mere formality?
- How will AI solving open problems at scale reshape the research division of labor and disciplinary ecosystem of human mathematicians?
Source and Original
This news item comes from TechCrunch AI (published on September 21, 2026, at 20:15:58). This site provides Chinese translations and commentary on overseas AI news; the original text is copyrighted by the original author.
Daily aggregation of overseas AI news and in-depth perspectives. Bookmark this site to never miss an important signal; Return to homepage for more.