OpenAI Announces Three New Ambitious Results in Mathematics and Physics
OpenAI has announced three new mathematical and physics results that it claims were achieved using artificial intelligence models. The company has launched a new publication plan where external experts will act as 'shepherds' to verify these findings within the scientific community.
OpenAI claims to have achieved a solution to the Exponential Time Hypothesis (ETH), a new proof for the Kakeya conjecture, and significant findings related to quantum field theories. The company announced that it has developed a new model to collaborate with external experts to present these results to the scientific world.
OpenAI has shared three new scientific developments that it states were obtained through artificial intelligence models. The company's claims include the refutation of the ETH, a solution to the Kakeya conjecture, and findings on quantum field theories involving a unified completion of the Standard Model. Some of the mathematical results were verified using the proof assistant Lean, which is used to reduce the margin of error.
The company stated that these findings are not seeking academic credit, describing the obtained data as 'raw research data.' OpenAI aims for external experts to take on a 'shepherd' role to form working groups in the process of transforming this data into scientific knowledge. However, the company acknowledges that this method could create information asymmetry.
Assessment: Following the 377 mathematical results previously shared by OpenAI, this new announcement continues the debate regarding the verifiability of results obtained through the company's closed models. It is noted that for some physics studies, the data has not yet been shared, and experts prefer to write these results themselves.
Sources
- gizmodo.com · Link