Artificial Intelligence
OpenAI forms mathematics advisory group
OpenAI said it had trained a new internal model, claimed major progress on open problems in mathematics, and said an independent advisory group would help guide how results are reviewed and shared.
- OpenAI said it began training a new internal model on 28 August.
- OpenAI said the model had resolved the Navier–Stokes Millennium Prize problem and more than 100 long-standing open problems.
- OpenAI said the pace of progress had surprised its mathematicians and led to internal discussions about how to inform the community.
- OpenAI said it was working with mathematicians who had formed an independent advisory group, hosted by the Institute for Advanced Study.
- OpenAI said the group would advise on review, communication, dissemination and standards, and would not be paid by OpenAI.
OpenAI said it began training a new internal model on 28 August and claimed the system had already resolved the Navier–Stokes Millennium Prize problem and more than 100 long-standing open problems. It said the speed of progress had surprised mathematicians inside the company.
The company said this led to internal discussions about how to tell the wider community about the work and how to prepare the field for it. It also said mathematics was a fundamental science with possible applications beyond the subject itself.
OpenAI said mathematicians had raised concerns in an open letter titled A Severe Misalignment of AI in Mathematics about using open problems as a benchmark for new AI systems. In response, it said it was working with mathematicians who had formed an independent advisory group on mathematics and artificial intelligence.
The group was described as being hosted by the Institute for Advanced Study and would include François Charles, Camillo De Lellis, Timothy Gowers, Martin Hairer, Nikhil Srivastava, Ulrike Tillmann, Ravi Vakil, Edward Witten and Melanie Matchett Wood. OpenAI said the group would review emerging results, advise on how they were shared, and help on academic and professional standards, while remaining independent and unpaid.
Named in this story
People
- François Charles
- is listed as an initial member of the advisory group
- Camillo De Lellis
- is listed as an initial member of the advisory group
- Timothy Gowers
- is listed as an initial member of the advisory group
- Martin Hairer
- is listed as an initial member of the advisory group
- Nikhil Srivastava
- is listed as an initial member of the advisory group
- Ulrike Tillmann
- is listed as an initial member of the advisory group
- Ravi Vakil
- is listed as an initial member of the advisory group
- Edward Witten
- is listed as an initial member of the advisory group
- Melanie Matchett Wood
- is listed as an initial member of the advisory group
Companies
- OpenAI
- said it had trained a new internal model and was forming the advisory group
Organisations
- Institute for Advanced Study
- hosts the advisory group
- Advisory Group on Mathematics and Artificial Intelligence
- will advise on review, communication and standards for emerging results
How the source tells it
The piece was self-promotional and self-justifying, pairing extraordinary claims about a model’s progress with repeated reassurance about responsible oversight.
- hype a company-authored claim of major mathematical breakthroughs is presented as settled progress
- reassurance repeated emphasis on an independent advisory group and responsible deployment softens the announcement