PSI Says Anthropic-Powered AI Solves Long-Standing Quantum Computing Problem
The research company claims an AI system using an unreleased Anthropic model solved the Aaronson–Ambainis conjecture, a major mathematical problem that had remained unresolved since 2008.
Physical Superintelligence Inc. (PSI) has announced that an artificial intelligence research system powered by an unreleased Anthropic model has produced a proof of the Aaronson–Ambainis conjecture, a long-standing problem in quantum computing. The company announced the achievement on November 6, 2026, claiming its system completed the proof within three days. The conjecture, originally developed by researchers Scott Aaronson and Andris Ambainis, addresses when quantum computers can achieve significant speed advantages over conventional computers.
According to PSI, the research system was developed internally and operated using an unnamed, non-public Anthropic AI model. The system reportedly generated a mathematical proof that was subsequently checked using Lean, a specialised proof assistant designed to verify mathematical arguments. PSI said the system later improved the result twice, with Lean accepting the revised versions. The project was led by David Turturean, a member of PSI’s technical staff, who developed the system, directed the research process and reviewed the resulting paper.
The Aaronson–Ambainis conjecture concerns the limitations of quantum computing, particularly whether quantum machines can achieve exponential speed advantages when processing random inputs. The reported proof suggests that such dramatic advantages generally require special patterns or structures in the information being processed. This does not mean quantum computing is ineffective, but rather that its greatest advantages may be limited to particular types of problems. Researchers familiar with the subject have described the reported achievement as a major development in theoretical computer science.
Despite the significance of the announcement, the proof has not yet undergone formal peer review, and independent experts are still examining the mathematical work. Researchers from Stanford University, including Guy Blanc, Jordan Docter, Carmen Strassle and Li-Yang Tan, have reportedly reviewed the argument and are preparing a more accessible explanation for specialists. Anthropic has not publicly identified the model used or independently confirmed its contribution. If validated, the achievement could represent another major milestone in AI-assisted scientific research, demonstrating how advanced AI systems may help researchers tackle problems that have challenged mathematicians for years.
