OpenAI Announces Three Breakthroughs in Scientific Research

A New Approach to Scientific Research

OpenAI has entered a new phase of research, announcing three fundamental discoveries across mathematics, physics, and computer science. Following last month’s progress in solving complex mathematical problems, the company has decided to involve independent experts in the dissemination of its research.

Notably, OpenAI has moved away from publishing results unilaterally. Instead, it has established groups of "shepherds" to oversee findings in computer science and physics. In the case of mathematical discoveries, researchers are being granted the opportunity to personally analyse the logic generated by the AI.

Mathematical and Computational Breakthroughs

In the field of mathematics, the AI has solved the "local smoothing conjecture" for Euclidean wave equations. This achievement automatically resolves complex mathematical tasks such as the Kakeya maximal and dimension conjectures. It is worth noting that just a few months ago, Professor Hong Wang was awarded the Fields Medal for a partial solution to the Kakeya conjecture, underscoring the scale of OpenAI's achievement.

In computer science, the company stated that its model has disproven the "Exponential Time Hypothesis" (ETH), a theory that has been considered true by scientists since 1999. The verification of these results is being led by leading experts in the field, including Russell Impagliazzo and Ramamohan Paturi, who were themselves the authors of the hypothesis.

Physics and Ethical Challenges

The most ambitious claim relates to the field of physics. OpenAI asserts that it has achieved a mathematically rigorous proof of quantum field theories, which includes the unification of quantum chromodynamics and the Standard Model. This work is being supervised by renowned physicist David Gross. However, the full data has not yet been made public.

Despite this progress, OpenAI's approach has faced criticism. An independent advisory panel has urged the company to refrain from the selective dissemination of results, as it creates risks of unequal access and bias. Experts recommend that such discoveries be placed in official scientific repositories and accompanied by formalised Lean proofs.

This move by OpenAI demonstrates that artificial intelligence is increasingly becoming an integral part of scientific research. However, how these discoveries will be integrated into academic circles, and the reliability of complex mathematical constructs created by AI, will remain one of the primary challenges of the coming years.