OpenAI creates a math advisory group after claiming 100-plus AI solutions
Navier–Stokes Millennium Prize problem
A famous math problem named in OpenAI’s announcement.
advisory group
A team that gives expert advice but does not make company decisions.
Fields Medal
A major international award for outstanding mathematicians.
What happened
OpenAI, the company behind ChatGPT, announced an independent advisory group on mathematics and artificial intelligence on September 21. The group is hosted at the Institute for Advanced Study in Princeton, New Jersey. Nine mathematicians are its initial members.
The announcement followed a major claim about an internal OpenAI model. OpenAI says the model began training on August 28. It says the model resolved the Navier–Stokes Millennium Prize problem. It also says the model resolved more than 100 long-standing open problems across most areas of mathematics. These remain claims from the company. The announcement does not establish independent acceptance of every result.
The background
Math is not finished when a machine prints a formula. Other mathematicians must inspect the reasoning, compare it with earlier work, and explain why it matters. They also need to know who contributed the underlying ideas. A fast answer can be useful, but a rushed announcement can make errors, credit disputes, and confusion harder to resolve.
Earlier in September, 25 Fields Medal winners signed an open letter. They warned that AI labs were racing to announce solutions to famous problems. Their concerns included missing explanations, weak links to earlier work, and the loss of human understanding that helps mathematics grow. The debate around OpenAI has also included questions about how researchers receive credit.
What the group will do
OpenAI says the group will assess the importance of new results. It will advise on how results are reviewed and communicated. It will also discuss academic and professional standards. Another focus is how AI tools can support mathematical research and learning.
The group is designed to operate independently. Its members will not be paid by OpenAI. They may give advice without being asked. They may publish their views. They can also control their own membership.
But the group has a clear limit. It cannot tell OpenAI to slow its internal math research. It cannot redirect that research. The Institute for Advanced Study also says it has no decision-making power over AI companies.
Why it matters
That boundary makes this more than a story about whether AI can solve math. It is a story about how knowledge enters a field. If AI produces many answers, someone must check them, credit earlier work, and turn them into explanations people can learn from. Counting solutions alone may reward speed without building understanding.
There is also a possible benefit. Mathematicians could choose important questions and use AI as a tool for exploration. They could then develop clearer proofs and new ideas. The advisory group is meant to help discuss that transition, not to control OpenAI’s research.
What is confirmed, and what remains unknown
Several facts are clear. OpenAI announced the group, named nine members, and described its independence, unpaid status, public voice, and limited role. The Institute for Advanced Study describes the group as a channel between mathematicians and AI companies. OpenAI also made the 100-plus claim.
Many important details remain open. The public does not yet have a complete list of the claimed problems. It is also unclear which proofs have been checked by outside experts. “Resolved” may not mean the same thing for every result. We do not yet know how much the group can change OpenAI’s release practices.
What to watch next
The next tests will be public proofs, independent checking, and clear credit for original researchers and collaborators. Watch whether the group publishes disagreement, not only advice. The larger question is whether AI progress will be judged by the number of answers, or by how well people can verify and understand them.
OpenAI says its AI solved over 100 math problems
📰 Full story: OpenAI creates a math advisory group after claiming 100-plus AI solutions
OpenAI says its AI solved more than 100 hard math problems. Now mathematicians will help explain and check what that means.
open problem
A math question without an accepted answer.
Fields Medal
A major award for excellent work in mathematics.
Institute for Advanced Study
A research center in Princeton, New Jersey.
💡 The gist
- OpenAI, the company behind ChatGPT, created a math advice team.
- OpenAI says one internal AI solved more than 100 open problems.
- The team can advise OpenAI, but cannot slow its research.
OpenAI announced the team on September 21. It is hosted by the Institute for Advanced Study, or IAS. IAS is a research center in Princeton, New Jersey. Nine mathematicians joined first.
OpenAI says an internal model began training on August 28. It says the model solved the Navier–Stokes Millennium Prize problem. It also says the model solved more than 100 long-standing open problems. An open problem is a math question without an accepted answer.
These statements come from OpenAI. They do not prove that every answer is accepted. Other mathematicians must read and check the work. They need to know whether each proof is complete. They also need time to understand new ideas.
Why did OpenAI create the team?
AI can produce answers very quickly. Math needs careful checking and clear explanations. It also needs fair credit. People need to know whose earlier ideas helped.
Earlier this month, 25 Fields Medal winners signed an open letter. They worried that AI companies were rushing to announce famous results. They also worried about explanations and credit. This debate is not only about speed. It is about how math grows and teaches people.
What can the team do?
The team can advise on new results. It can discuss their importance. It can advise on how OpenAI shares them. It can discuss research standards. It can suggest ways to use AI for math research and learning.
The members will not receive pay from OpenAI. They can give advice without being asked. They can publish their opinions. They can choose their own future members.
What can it not do?
The team cannot tell OpenAI to slow its math work. It cannot redirect the company’s internal research. IAS also says the team has no decision power over AI companies. It is an advice team, not a governing body.
What should we watch?
We still do not know every problem in the claimed 100-plus results. We do not know which proofs outside experts have checked. We also do not know what OpenAI means by “solved” for every result.
Next, watch for public proofs and independent reviews. Watch how researchers receive credit. The important measure may become understanding, not just speed.
A math helper team for a very fast AI
📰 Full story: OpenAI creates a math advisory group after claiming 100-plus AI solutions
OpenAI says its AI solved many hard math puzzles.
mathematician
A person who studies numbers and patterns.
math puzzle
A hard math question waiting for an answer.
AI
A computer helper that finds answers.
A simple story
OpenAI, a company that makes AI, built a math team. AI is a computer helper that finds answers. OpenAI says its AI solved more than 100 math puzzles. These puzzles had no known answer before. Nine mathematicians joined the team. A mathematician studies numbers and patterns. They will read the AI’s answers. They will help people understand the answers. But they cannot make OpenAI slow down. We still do not know every puzzle. People must check the answers carefully. Fast answers are not enough. People must understand the work too.