The Mathematics Community and AI
OpenAI has demonstrated a pattern of making significant mathematical breakthroughs while struggling with the communication and public rollout of these achievements. To address fractured relations with the mathematical community, the company has begun consulting a newly formed independent advisory group.
OpenAI keeps bulldozing mathematicians
The Verge spoke to one of the mathematicians advising OpenAI as the field braces for a new wave of results that could upend their work.
OpenAI keeps bulldozing mathematicians
The Verge spoke to one of the mathematicians advising OpenAI as the field braces for a new wave of results that could upend their work.
OpenAI’s latest attempt to repair relations with the mathematical community involves consulting an independent advisory group of elite practitioners. However, mathematicians describe a messy rollout that bears the hallmarks of previous rushed announcements. The newly formed group faces the daunting task of helping coordinate the release of scores of additional results produced by an unreleased model, raising concerns among researchers about the impact on their field.
“The way they phrased their announcement didn’t really help.”
Formation of the Advisory Group
An assortment of eminent mathematicians announced the formation of the Advisory Group on Mathematics and Artificial Intelligence (AGMAI) via a blog post by UCLA mathematics professor Terence Tao. The independent body consists of nine elite mathematicians intended to advise frontier AI labs on the review and communication of emerging results. OpenAI stressed that the group operates independently, with members free to challenge the company publicly, while noting the group will not dictate internal research pacing.
But many mathematicians, and certainly most people outside of that community, learned about that group only through OpenAI’s much louder rollout.
The rollout generated widespread confusion regarding whether AGMAI was truly independent or an OpenAI-appointed body. AGMAI's website notes that the group formed after OpenAI approached potential members about establishing an advisory board before they decided to organize independently.
Martin Hairer, an AGMAI member and mathematics professor at Imperial College London and EPFL, emphasized the group's complete independence from OpenAI. He stated that the group receives no financial or technical support from the company and that members have not signed restrictive agreements beyond standard confidentiality requirements for advanced research access.
Hairer described the preceding days as an intense experience, noting that the group is independent, unpaid by OpenAI, and open to working with other frontier AI labs.
Solutions are often less important than the understanding that comes with them. In human mathematics, the two have traditionally gone hand in hand, and AI seems to be changing that.
Hairer expressed frustration regarding how OpenAI handled the announcement. Given past controversies surrounding credit and the treatment of mathematicians following the solution of a Millennium Prize problem, the community remains naturally suspicious of corporate spin.
AGMAI must win the trust of mathematicians to help navigate an increasingly fraught relationship with AI companies. Hairer acknowledged that OpenAI's wording was careful but unhelpful in establishing immediate credibility for the fledgling advisory group.
The entire endeavor came together rapidly, with Hairer noting that the group was still in its formative stages without a definitive master plan.
Upcoming Results and Academic Concerns
The urgency stems from OpenAI holding a tranche of results ready for release. Following the resolution of the Navier-Stokes Millennium Prize problem, OpenAI reports that its unreleased internal model has resolved more than 100 long-standing open problems across various areas of mathematics.
While assembling a team of top mathematicians is intended to avoid public relations disasters, trumpeting a large volume of pending results while promising responsible handling highlights OpenAI's position as an outsider to standard academic practices.
Álvaro Lozano-Robledo, a mathematics professor at the University of Connecticut, criticized the company's approach, stating that academic researchers do not typically announce unreleased proofs in such a manner.
Lozano-Robledo noted a disconnect in other corporate pronouncements regarding Millennium Prize problems, emphasizing that academic mathematicians view problems as either solved or unsolved.
Researchers have raised concerns over poorly written manuscripts, scant engagement with existing literature, and undocumented modifications to released documents. Critics characterized these practices as sloppy scholarship that breeds paranoia within the research community.
Lozano-Robledo described the influx of AI-generated math solutions as a burden on the community, noting that automated solutions lack the deep conceptual understanding traditionally tied to mathematical discovery.
“So we, well, you know, started two days ago. So it’s not like we have a big master plan.”
Despite producing impressive computational results, AI systems still rely on human mathematicians to evaluate significance and contextualize findings.
Impact on Researchers
For many researchers, the uncertainty surrounding upcoming announcements creates a tense atmosphere. The prospect of OpenAI releasing over 100 results without clear details generates anxiety that years of dedicated academic work could be unexpectedly disrupted.
Lozano-Robledo accused the company of fueling a frenzy and increasing dread among mathematicians whose specific areas of study might be impacted by future announcements.
Colva Roney-Dougal, a mathematics professor at the University of St Andrews, noted that the impending announcement of a large number of results creates professional uncertainty and anxiety regarding the potential obsolescence of ongoing student and faculty research.
This rapid pace leaves researchers at a scholarly crossroads, unsure whether to accelerate publication efforts or wait for upcoming corporate disclosures.
Hairer acknowledged the anxiety felt across the field, noting that potential disruption to researchers motivated him to join efforts addressing the misaligned goals of AI companies and the mathematical community.
While acknowledging potential public relations spin by AI companies, Hairer emphasized that his primary concern remains the well-being of the mathematical community as a whole.
