From Futurism · Frank Landymore · · 1m 10s
OpenAI's math breakthrough is unreadable
OpenAI's AI agents solved a legendary math problem, but mathematicians can't understand the proof.
In brief
OpenAI's AI agents solved a legendary math problem, but mathematicians can't understand the proof. AI can solve hard problems faster than humans, but the solutions may not advance human knowledge if we can't comprehend them. Originally reported by Futurism.
OpenAI claims math breakthrough
OpenAI announced its AI agents solved the Navier-Stokes equations, one of math's biggest unsolved problems, after 88 hours of computation using 10,000 agents.
The proof is borderline incomprehensible
While technically correct, the AI's proof is so difficult to parse that mathematicians say it teaches them almost nothing about the problem.
“The paper is not written for humans.”
Plagiarism accusations emerge
NYU mathematician Tristan Buckmaster accused OpenAI of stealing his unpublished work on the same problem, claiming the company may have accessed his notes through its Codex tool.
OpenAI solved the wrong version
OpenAI's solution used an external force applied to fluids, but mathematicians are chiefly interested in the purer scenario without external forces. The company may not have solved the actual Millennium Prize problem.
The real problem remains unsolved
A recent proof by mathematicians contends that OpenAI's approach can never lead to solving the full Navier-Stokes problem. The core mystery of fluid dynamics persists.
“The most important problem is unsolved.”
In short
AI can solve hard problems faster than humans, but the solutions may not advance human knowledge if we can't comprehend them.
