AI at PSI helped solve an 18-year quantum computing problem — TechCrunch
David Turturian, an employee at the startup lab Physical Superintelligence (PSI) who recently completed his studies at the Massachusetts Institute of Technology, used early access to Anthropic’s Fable 5.2 model to solve an 18-year-old problem in quantum computing theory. It concerns the Aaronson–Ambainis conjecture, which relates to the ability of quantum computers to solve tasks faster than classical ones, TechCrunch reports.
Verification of the mathematical proof
One of the conjecture’s authors, scientist Scott Aaronson, wrote that he does not fully understand the resulting proof, but considers it correct. According to him, the proof was automatically formalized in Lean, a language used for mathematical proofs. Aaronson also noted that the result was explained by a group of leading researchers with whom PSI shared the solution.
The result does not rule out all possible ways quantum computers could accelerate the solution of complex tasks. At the same time, it closes off one line of inquiry. Turturian called it a negative result that allows researchers to no longer focus on the corresponding approach.
PSI’s business model
PSI CEO Matt Pines said that the company is creating specialized tools for scientific research, including memory systems and simulation infrastructure. According to him, such systems should make it possible to analyze a problem broadly and in detail, considering many possible options.
More current news is available on the UA.News Telegram channel Telegram.
PSI currently generates revenue by applying its tools to optimize data centers, where heat, electricity, water and noise metrics are modeled. In the future, the company expects to turn breakthroughs in theoretical physics into technologies.
Pines also noted that physics differs from mathematics because computational conclusions must be verified through experiments. PSI is focused on how to confirm computational results experimentally as efficiently as possible. The result concerning the Aaronson–Ambainis conjecture was not disclosed until Anthropic’s model approached release, giving PSI time to show the solution to specialists in the field.
PSI adviser and Northeastern University professor Jim Halverson noted that best practices for AI-assisted science are still taking shape. In his view, transparency and consultations with experts are important.
