Lifestyle
Altman sympathizes with mathematicians, but says OpenAI won't stop releasing AI math results
According to The Verge, OpenAI CEO Sam Altman says he sympathizes with mathematicians struggling to keep up with the company's AI math results, but it won't stop. He says it has more unpublished discoveries, as researchers criticize how it shares its work.
About 5 min read
What happened
The Verge reports that OpenAI CEO Sam Altman said on the Cheeky Pint podcast, released Thursday, that he sympathizes with mathematicians swamped by AI-generated results, but has no plans to stop.
As quoted by The Verge, Altman said: "I think, on net, the world deserves to understand how the universe works," and added, "But I do have empathy for the method." He argued that the public's right to new scientific knowledge outweighs the disruption caused by how the results are released, while acknowledging that the company needs to improve how it communicates.
As quoted by The Verge, Altman also said: "Humanity deserves fundamental knowledge, and we should not be sitting on a major fundamental science discovery."
Researchers' criticism and the controversy
According to The Verge, the mathematicians it interviewed repeatedly criticized how OpenAI releases its results, saying it does not make credit attribution clear, lacks context and does not verify enough. Researchers say they end up spending a great deal of effort checking and understanding the results themselves while the company keeps putting out new ones. The Verge says OpenAI has released hundreds of math results this month, often with only short explanations that researchers describe as poorly written and hard to follow.
On the other hand, The Verge notes that OpenAI has also been criticized for holding back important results. OpenAI announced what it claims are three major breakthroughs in mathematics, physics and computer science. Details of one claimed result, involving Yang-Mills theory (the mathematical foundation of modern particle physics), remain unpublished. According to the report, OpenAI says this follows the advice of a group of experts it selected and consulted. The Verge points out that this runs directly counter to the advice of another group of mathematicians the company consulted earlier.
The scale and cost Altman disclosed
According to The Verge, asked whether the company has more results beyond the hundreds of proofs already published, Altman answered, "We do. Of course." He also acknowledged that the company's pursuit of math problems did "get out of control" for a short period, but said this was deliberate and necessary to test the limits of its models. He mentioned an internal Slack (workplace messaging app) channel called "Model Name-Hold My Beer," where employees threw the models at the hardest problems they could.
The Verge reports that Altman estimates OpenAI has spent hundreds of millions of dollars on math problems. Of that, its solution to the Navier-Stokes problem cost about $10 million, and he said "maybe a little more." The Navier-Stokes problem is one of mathematics' famous "Millennium Prize Problems." The report notes that AI companies investing this many resources leaves some researchers worried they cannot afford to compete and may be pushed out of their own fields.
| Aspect | Altman / OpenAI's position | Researchers' position (per The Verge) |
|---|---|---|
| Release pace | Will not stop; more results remain unpublished | Hundreds of results this month, hard to absorb |
| Communication | Admits it needs to improve | Credit attribution, context and verification all insufficient |
| Yang-Mills result | Details withheld on expert advice | Criticized as holding back research |
| Resources | Estimated hundreds of millions of dollars | Fear being pushed out of their fields |
What happens next
According to The Verge, Altman does not plan to slow down and hinted that more results remain unpublished. The Verge suggests that mathematicians and researchers hoping for a pause in AI math releases to catch their breath will likely be disappointed.
Frequently asked questions
What did Altman say?
According to The Verge, he said he sympathizes with mathematicians but believes the world deserves to understand how the universe works, so OpenAI will not stop releasing results; he also acknowledged the company needs to improve how it communicates.
Does OpenAI have more unpublished results?
According to The Verge, when asked whether there are more beyond the hundreds of proofs already published, Altman answered, "We do. Of course."
What are researchers mainly criticizing?
According to The Verge, the criticisms include unclear credit attribution, missing context, insufficient verification, and unpublished details for some results (such as the Yang-Mills-related one).
Where do the spending figures come from?
They are Altman's own estimates on the podcast, relayed by The Verge: hundreds of millions of dollars in total, with about $10 million for the Navier-Stokes solution, "maybe a little more."
Browse the latest news in this topic
Lifestyle
DoorDash: Third-Party AI Agents Have Brought Only "Several Hundred" Orders So Far, and Most People Still Order in the App
According to TechCrunch, delivery company DoorDash said on its third-quarter earnings call that third-party AI agents that can place orders on users' behalf have brought in only about several hundred orders. DoorDash's 2025 report says it completes an average of 3 million deliveries a day, so the figure is tiny by comparison.
Lifestyle
AI Company Cohere Expands Canada Commitment: Over C$4 Billion and 1,500+ Jobs Expected by 2031
Toronto-founded AI company Cohere says it will put more than C$4 billion into research and development (R&D) and AI infrastructure in Canada from 2024 to 2031 and add over 1,500 jobs. The plan matters most to students, tech talent and businesses adopting AI in Canada, but the investment and job figures are the company's own projections.
Lifestyle
Startup Runware Wants to Mass-Produce AI Data Centers Like Cars: The 48-Hour Goal and the Risks
TechCrunch reports that Runware plans to build AI data center pods on a production line, ready to ship as a unit. If it works, companies that need AI computing power could set it up faster, but these are still plans, and funding, supply chain and execution remain untested.
Lifestyle
AI Data Center Backlash in Maine Becomes a Senate Race Flashpoint, The Verge Reports
The Verge reports that many Maine residents worry AI data centers raise power bills, drain resources and bring few lasting jobs. That backlash has become a Senate race issue, and the result will help decide which party controls the Senate and shape US data center and AI policy.