OpenAI reports research advances in mathematics and physics

OpenAI announced that its AI models contributed to discoveries in harmonic analysis, computational complexity and quantum field theory through collaborations with academic scientists. The work focused on theoretical research, using models to support hypothesis generation and identify connections across disciplines rather than autonomously conduct experiments.

In harmonic analysis, AI helped formulate a conjecture about neural network structure and approximation capabilities. In computational complexity, it assisted in discovering a more efficient matrix multiplication algorithm, although OpenAI did not quantify the improvement over the previous best-known approach. In quantum field theory, the models helped develop a method for interpreting and computing properties of certain theoretical particles and fields.

Researchers used large language models to review scientific literature, cross-reference concepts and explore complex problems. Human experts guided the process and validated the outputs, keeping the emphasis on augmenting scientific judgment rather than replacing it. Detailed findings are available in academic preprints, but applying the algorithm or physics methods requires substantial domain expertise. The internal tools used for the research are not public, limiting immediate adoption beyond specialized research teams.