NoorSadaNoorSada
BREAKINGEditorialDeveloping✓ 1 source

OpenAI says AI models proved long-standing maths conjecture and two other scientific results

The company's models resolved the local smoothing conjecture in harmonic analysis, disproved the Exponential Time Hypothesis, and established the existence of quantum chromodynamics.

Speed:

PD
Pierre Dubois
· 2 min read
The Pioneer Building in San Francisco houses OpenAI's offices · Photo: HaeB / Wikimedia Commons (CC BY-SA 4.0)

OpenAI said Monday its AI models had produced scientific results in mathematics, computer science and physics, including the first proof of the local smoothing conjecture across every spatial dimension.

The conjecture, a central proposition in harmonic analysis — the study of how complex signals break down into elementary waves — had previously been proved only in three dimensions. Alex Iosevich, a professor of mathematics at the University of Rochester who specialises in the field, called it a historic development, saying he was filled with joy and optimism at the news.

The AI models also disproved the Exponential Time Hypothesis, which concerns the time required to solve complex computational problems, and provided mathematically rigorous proof of the existence of quantum chromodynamics, the theory governing the interaction of quarks and gluons.

OpenAI, valued at $852 billion, said it reached out to scientists outside the company to help interpret the raw research data. It used three expert groups, which it called shepherds, to advise on how to release findings.

The groups working on mathematics and computer science chose to release their data publicly immediately. The physics group decided to delay publication in order to write their own interpretation first.

Iosevich said the field of theoretical mathematics would change drastically as a result of AI, adding that many mathematicians he knew shared a twinge of sadness alongside their optimism. He said problems in pure mathematics had often emerged from concrete physical considerations, and those connections could now be explored rapidly.

Sources — how we know this