Explained: OpenAI solves 90-year-old maths problem humans could not, so why is everyone angry at Sam Altman

OpenAI says ChatGPT solved a very hard maths problem, but mathematicians are not buying it.OpenAI has solved the Navier-Stokes problem. Days after launching GPT-6 Astra, which veterans including Nvidia CEO Jensen Huang have publicly described as AGI, OpenAI has begun teasing an unannounced model that it says is even stronger at solving very hard mathematics problems. The company has claimed that the model solved the Navier-Stokes problem, a problem that has challenged humans for about 90 years and is one of the seven Millennium Prize problems.Given the hype and buzz around the Navier-Stokes problem, curiosity is but expected. Even more so because controversy is known to follow OpenAI and its chief Sam Altman. This announcement, too, has quickly become controversial, with mathematicians raising questions about whether OpenAI relied on work by others—humans—to solve the problem. Before getting into the controversy itself, it is imperative to understand what OpenAI has solved or rather, what is this Navier-Stokes problem?What has OpenAI solved and how?The Navier-Stokes is a family of equations related to fluids. More precisely, it is a mathematical rulebook that uses Newton's laws to predict how fluids—think water and gases—move in space. Measurable metrics like velocity and pressure are used to make predictions. The problem seeks to find if a perfectly normal fluid flowing naturally can break itself. In other words, can the math break down and cause the fluid's velocity to reach infinity aka singularity over a finite amount of time.The problem — or rather a bunch of problems which all centre around Navier-Stokes algorithms — is centuries old, although over time different parts and bits of it have been solved by physicists and mathematicians. The most famous of the equations that remain unsolved is the one that is part of the seven Millennium Challenges created in the year 2000. OpenAI did the math. 10,000 OpenAI agents, controlled by an unreleased AI model that is supposedly next level and magnitude more intelligent than the just launched ChatGPT Astra, were put to task. They worked for 88 hours. The operation cost millions in AI tokens and compute. Actually, the agents were divided into large teams and these AI agent teams then pursued the solution in parallel. In total they generated over 300 billion tokens and coordinated via 4.9 million messages with each other. The team that ultimately cracked the problem used over 130 billion tokens.Call it brute force or AI intelligence, eventually the code was cracked. As per the company, its AI agents navigated endless complex simulations and found a specific mathematical scenario—a fluid vortex—where the flow breaks down spiralling inward, stretching thinner and shrinking smaller until the speed in its exact centre hit infinity.Why it is a big deal? The big deal here is that OpenAI agents were able to reach this stage without external guidance or inputs from human engineers. Or so says OpenAI. The fluid's internal forces (pressure, acceleration, viscosity) all went to extremes but perfectly cancelled each other out, allowing the singularity to form naturally. The solution would ordinarily be worth a $1 million reward, although OpenAI says it does not plan to take the prize.The world, as OpenAI announced its solution, seems to have turned upside down. There are implications. And then there are some. Different people are reacting to the news in a different way. Some are calling it the end of mathematics. Some are saying that what humans could not solve has finally been solved by AI and that means the era of humans being the most intelligent entity on the earth is over. Some are saying that a super-intelligent AI coordinating over 10,000 agents to crack a challenge humans could not for decades mean that AI has achieved an ability that earlier only humans had — that is effectively strategising and coordinating to solve something that seems insurmountable.Whether any or all of these assertions are accurate, one thing is certain: the field of pure scientific research has changed. Terrance Tao, a celebrated mathematician and winner of Fields Medal, summed it up in a post on social media. “We have now seen that even the rumour of someone working on a problem can trigger a massive amount of AI-powered effort to flatten it before the original research project has time to reach its full potential,” he wrote. “The incentives may now be pointing in the direction of no longer sharing any promising research directions with the broader community, which would reverse centuries of traditions of open science and do serious long-term damage to the future of the field.”And this brings us to the point about the world hating on OpenAI and Sam Altman due to how the company has gone about solving the Navier-Stokes problem. You see, there has been a lot of drama, heartburn and finger-pointing in the last few days.Why are mathematicians angry?In the last few days, New York University mathematics professor Tristan Buckmaster has alleged OpenAI moved quickly on the problem after learning of work by him and Levent Alpoge, a researcher from rival AI company Anthropic. Buckmaster has also alleged that OpenAI tried to influence who should receive credit for solving the problem.OpenAI, on its part, says the company began training a new AI model with advanced mathematical capabilities on August 28. That is because it heard rumours that its arch rival Anthropic was making progress on Navier-Stokes. OpenAI decided to commit more resources to the effort, presumably in a bid to beat Dario Amodei.Soon after, Alpoge and Buckmaster posted documents claiming key advances in an area related to the Navier-Stokes problem. They said they used several AI models, including Claude and Codex (which is made by OpenAI), in their work. Buckmaster later said that last week he learned OpenAI had become aware of their work and had begun committing significant resources to the problem. He said he asked OpenAI leaders whether the company had accessed the pair's Codex logs, and was told the model didn't look up user data, though he claimed the company did not answer questions about training.Buckmaster further said OpenAI made several proposals, including one under which he could publish a paper saying the Navier-Stokes problem had been solved by an internal OpenAI model without Alpoge's name being included.Sebastien Bubeck, a member of technical staff at OpenAI rejected the suggestion that OpenAI had proposed removing Alpoge's name from credit. OpenAI chief executive Sam Altman also defended the team's work, saying, “Seb—and everyone else—acted with integrity and generosity throughout. It is true that we tried this because there were rumors on the internet last week that Anthropic's models had solved a millennium problem, and we were curious if ours could do it too.”Buckmaster or Alpoge haven’t directly accused OpenAI of stealing their work to solve the problem. OpenAI, even if it doesn’t say that their Codex logs might have been used, says, “we cannot rule out that de-identified data derived from their usage of our products helped improve our models.” In other words, while someone did solve a 90-year-old math problem, whether it was solved by a human or AI remains open to interpretation. Sam Altman on his part says OpenAI’s latest model can solve “many, many other math problems.”For now, not everyone believes him and mathematicians in particular seem quite mad at his company for doing the research the way it did, even if it resulted in a solution to a significant problem.- EndsPublished On: Sep 9, 2026 16:43 IST