Surreal! A 18-year-old open question regarding whether a quantum computer can be exponentially faster has now been demonstrated as a proven result. An artificial intelligence system wrote the mathematical proof, and on the third day, Lean—a program that verifies such texts—fully accepted it. Physical Superintelligence, a research company, says its internal system generated this proof using a non-public Anthropic model. The company refined the proof twice and published the files on a public page. The text states that all four main proofs are complete, with no blank sections or extra rules, and that the result has been re-verified by two independent versions of Lean. The conjecture was originally posed by Scott Aaronson and Andris Ambainis, questioning whether quantum computers can achieve that speedup even when the problem is random and lacks structure. According to the company’s page, a Stanford team (Guy Blanc, Jordan Docter, Carmen Strassle, and Li-Yang Tan) called this a monumental result that resolves a decades-old conjecture; in August, their own paper had only addressed a special case. https://t.co/UoekWZgylC