The announcement and the role of AI
Physical Superintelligence (PSI), a research institute cited by the press, has announced that it has proved the Aaronson-Ambainis conjecture, an open problem since 2008 concerning the conditions under which a quantum computer can achieve a computational advantage over a traditional computer. According to reports, the proof was produced by a research system developed internally by PSI and based on an Anthropic model not yet released, with a stated working time of three days.
The new element is not only the mathematical result claimed, but the method: a system built for AI-assisted research, attributed to a PSI technician, David Toutourian, who reportedly used an unidentified model. The news, picked up by IT之家 from a Business Insider report, does not indicate which model was used nor provide the text of the proof, something that makes independent verification of the result impossible.
What changes and what does not
The content of the conjecture, as reported, clarifies a relevant theoretical point: the exponential advantage of quantum computing would not be automatic, but would require the presence of specific structure or regularity in the problem being addressed. The proof, if confirmed, would support the hypothesis, long discussed among researchers, that quantum computers are not superior to classical computers in every scenario, while retaining strong advantages on certain classes of problems.
This second aspect is important so as not to overestimate the news: it is not the end of quantum computing, but the delimitation of the conditions under which it can express its advantage. The same researchers who had studied the problem reportedly described the result as a significant milestone, according to the account provided. However, it remains a reported claim, not an independently verified fact.
The material also stresses that quantum computers already exist but the technology is still at an early stage: a contextual element useful for situating the practical scope of the discovery, which for now remains theoretical.
The context and points to be verified
The announcement fits into a sequence of cases in which AI is presented as a tool for mathematical research: in September of the same year, according to the material, OpenAI had announced the solution of the Navier-Stokes equation problem, which had remained open for ninety years. The comparison between the two episodes shows a common communications tendency, but also a substantial difference: in the PSI case, the model used is not public and is not named, which limits the replicability of the work.
Among the open questions are the identity of the Anthropic model used, the availability of the research system developed by PSI, the existence of a peer-reviewed publication, and independent confirmation of the proof. Until these elements are available, the news should be treated as an announcement by the institute, reported by the press, and not as an established scientific result.
