The Math That OpenAI Keeps for Itself? AI Has Solved a 90-Year-Old Conundrum

AI Summary

OpenAI is causing a stir in the academic world by claiming to have solved a difficult math problem using a powerful, unreleased AI model.

Imagine this: There is a puzzle that has plagued the world’s greatest mathematicians for decades. What if a mysterious “digital entity” suddenly solved this problem—which had remained a mystery for 90 years—in just a few days?

This is exactly what has been happening in the field of artificial intelligence recently. OpenAI announced that it has solved a difficult mathematical problem using an unreleased AI model that has no name, no price, and no public API. The mathematical community is currently in an uproar. Is this a landmark event in the history of human mathematics, or is it a controversy hidden behind technological monopoly?

Why is this important?

Mathematics is the “root” of all science and technology. The “Navier-Stokes equation” problem that OpenAI tackled is a high-level physical formula that explains the flow of fluids—how air moves and how water flows. Completely understanding and solving this equation would provide the foundation for a massive technological leap that could fundamentally improve human life, from optimizing aircraft design to precise weather forecasting.

However, the real reason this announcement is drawing the attention of both academia and the public lies not in the technological achievement itself, but in the “method.” The AI model that solved a conundrum humans have struggled with for decades in just 88 hours is an “OpenAI secret weapon” that has never been released to the world. Because the results were generated inside a corporate “black box” rather than being transparently verifiable by anyone, a fierce debate continues regarding its impact and the validity of the proof.

Easy to understand: Spaghetti created by 10,000 ‘AI clones’

How was this incredible feat possible? OpenAI mobilized approximately 10,000 “AI agents” (AI systems that carry out tasks autonomously) simultaneously to solve this problem. Source 1

To use a simple analogy: imagine there is a massive puzzle that cannot possibly be solved alone. OpenAI created 10,000 “AI clones,” assigned each of them one piece of the puzzle, and had them solve it simultaneously. Source 3 They cooperated with each other, continuously building logic for 88 hours. OpenAI described the structure of the equation created in this process as being like “spaghetti” swirling and curling inward. Source 6

Of course, there are question marks attached to this method. Scientific publications raise the possibility that in the process of solving the Navier-Stokes problem, OpenAI may have arbitrarily applied external conditions, thereby slightly modifying the problem itself. Source 10 In other words, there is suspicion that it may have solved a form somewhat distant from the problem originally defined by mathematicians.

Current situation: The ‘End of Math’ or an ‘Unfinished Task’?

On October 6, OpenAI released 722 math papers written in this manner. Source 4 One computer scientist even assessed it as “one of the most important days in the history of mathematics.” Source 7

However, reality is not so simple. While some claim these proofs were verified through “Lean” (a programming language for verifying mathematical proofs), the papers are still classified as “unchecked” in OpenAI’s official records. Source 2 Ultimately, the academic world is still in the stage of carefully verifying whether these results are truly mathematically perfect “truth” or merely elaborate fantasies created by AI. Source 11

What happens next?

This incident is a powerful signal that AI has deeply penetrated the realm of mathematical thinking. Although the resolution of the Navier-Stokes problem will not immediately change airplanes or weather forecasts, the ability of AI to delve into the foundations of mathematics is now exceeding human imagination. Source 5

What we must focus on in the future is the “monopoly” of AI technology. No matter how advanced a model may be, if the technology is strictly kept as a corporate secret, joint verification by scholars around the world becomes difficult. We must watch to see if OpenAI reveals the identity of this “secret model” to the world and begins true academic exchange, or if it continues to develop while holding a monopoly on this “sanctuary of mathematics.”

AI’s Take

It is a remarkable advancement for AI to solve human conundrums. However, mathematics is a field where the process is more important than the result. No matter how genius an AI may be, if it does not undergo transparent verification by the academic community, its answer remains a “corporate claim,” not “truth.”