Research

PSI says AI system proved Aaronson-Ambainis conjecture; paper not yet peer reviewed

Physical Superintelligence (PSI) announced on Nov. 6 that a research system it built had proved the Aaronson-Ambainis conjecture, posed by Scott Aaronson and Andris Ambainis in 2008. PSI says the system ran on its own infrastructure, used multiple instances of an undisclosed Anthropic model, and produced a complete proof on its third day of operation. The proof passed verification in the Lean proof assistant, and the paper and Lean formalization are on GitHub, but the paper has not undergone journal review. David Turturean led the project and takes responsibility for the paper's correctness, adviser Jim Halverson reviewed the proof, and Stanford researchers who proved a special case in August are reading the full proof.

Why it mattersIf the claimed proof holds, quantum computers could gain at most polynomial speedups over classical ones on random inputs, which helps define the limits of quantum advantage for unstructured problems. It would not remove the relevance of structure-dependent speedups such as Shor's 1994 factoring algorithm.

Read the full story at TOKENPOST →

Summarized by AI from the linked article.