OpenAI Claims Proof Of Three Scientific Problems
The San Francisco company says its models proved a long-standing math conjecture and disproved a complexity theory hypothesis.
By Rafael Collins ·
Artificial intelligence is moving from writing code to producing proofs, according to an announcement from OpenAI on Monday. The company said its models generated three new findings across mathematics, computer science and physics, covering harmonic analysis, complexity theory and quantum field theory.
In mathematics, OpenAI said its models proved the local smoothing conjecture for every spatial dimension. Researchers had previously established the result only in three dimensions. Alex Iosevich, a professor of mathematics at the University of Rochester who works in harmonic analysis, said OpenAI appeared to have "cracked" the conjecture in a "historic development."
Iosevich noted the conjecture was formulated by his PhD adviser, Chris Sogge. "The news was a shock at first," he said. He added that theoretical mathematics would "change drastically" because of AI. "I feel a twinge of sadness as a result," he said.
On the computer science side, OpenAI said its models disproved the Exponential Time Hypothesis, which concerns the time required to solve complex computational problems. In physics, the company said it produced a "mathematically rigorous [proof of the] existence" of quantum chromodynamics, the theory describing the interaction of quarks and gluons.
The ChatGPT creator said it consulted with scientists outside the company to turn its research data into scientific understanding. OpenAI selected three expert groups to act as "shepherds" for the findings. The groups handling the mathematics and computer science results released the AI-generated data immediately. The physics group chose to delay publication so its members could write their own interpretation of the results.
OpenAI said it weighed the tradeoffs. "All of these approaches involve tradeoffs," the company said. "On the one hand, the collaborations empower experts to foster unpacking the results and contextualizing them in the broader academic community. On the other hand, choosing initial shepherds necessarily shapes who gets involved first and creates asymmetry of information."
The San Francisco-based company, valued at $852 billion, invited feedback on expanding participation in the process.
The announcement follows other AI-assisted research claims. Last month, Anthropic, the creator of Claude, announced the discovery of a new enzyme said to resemble CRISPR, the gene editing mechanism used to treat conditions such as cancer and sickle cell disease. Scientists responded to that claim with mixed reactions, with some describing it as an exciting development and others suggesting the AI company had exaggerated its significance.
