From The Verge · Emma Roth · · 1 min
OpenAI's Math Breakthrough Ignites Data Drama
OpenAI claims to have solved the 90-year-old Navier-Stokes problem, but rival researchers suspect their private AI sessions fueled the proof.
In brief
OpenAI claims to have solved the 90-year-old Navier-Stokes problem, but rival researchers suspect their private AI sessions fueled the proof. AI breakthroughs in complex fields like mathematics are increasingly entangled with questions over training data provenance and user privacy. Originally reported by The Verge.
OpenAI claims math breakthrough
OpenAI announced it discovered a solution to the 90-year-old Navier-Stokes problem using a powerful internal AI model running alongside 10,000 concurrent agents.
“discovered a solution to the Navier-Stokes problem — which relates to the flow of liquid and gas”
Rivals release related work first
Just one day prior, NYU professor Tristan Buckmaster and Anthropic researcher Levent Alpöge published findings on a closely related mathematical problem.
“published findings on a related problem in partnership with Levent Alpöge, a researcher at Anthropic”
Suspicions over private code sessions
Buckmaster raised alarms after discovering OpenAI used a similar approach to his team's work, which had been drafted inside OpenAI's Codex tool.
“I asked whether the model had been trained on, or had access to, our sessions in Codex, into which we had been putting all our drafts”
OpenAI defends its training data
OpenAI denied accessing specific user data for the solution, though it admitted it could not rule out that de-identified user data influenced model training.
“while unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models”
OpenAI rejects $1M prize
OpenAI researcher Sebastien Bubeck insisted their proof differs significantly from Buckmaster's. OpenAI also confirmed it will not accept the $1 million Millennium Prize reward.
“One can in hindsight see that our proofs differ significantly and even the precise results proved are different.”
The essence
AI breakthroughs in complex fields like mathematics are increasingly entangled with questions over training data provenance and user privacy.
