PSI Claims to Have Proven Aaronson-Ambainis with an Anthropic Model

Physical Superintelligence Inc. announces that it has solved an open problem in quantum computing that has persisted since 2008, relying on an unpublished model from Anthropic. The lab claims a rapid demonstration and suggests that the quantum advantage would depend on particular structures. Researchers are praising the result, while Anthropic has not commented.

PSI Places Limits on Quantum Advantage and Receives Support

PSI indicates that the demonstration shows that certain exponential advantages of quantum computing only occur in the presence of specific patterns or structures. According to the lab, this means that quantum computers have limitations, a long-discussed hypothesis, even though they remain highly effective for certain tasks. Researchers who have worked on this problem describe the result as "huge," believing it resolves a decades-old question. Guy Blanc, Jordan Docter, Carmen Strassle, and Li-Yang Tan from Stanford University praised the proof where "everything comes together in a rather magical way," according to PSI's announcement. The precise identity of the Anthropic model used is not known, and Anthropic has not responded to requests for comment. Furthermore, quantum computers already exist but are still in an early stage of development.

Three Days of Computation and an Unpublished Anthropic Model, According to the Lab

PSI claims to have proven the Aaronson-Ambainis conjecture in three days, using a research system that incorporates an unpublished model from Anthropic. David Turturean, who is part of PSI's technical team, is the originator of this research system. The Aaronson-Ambainis conjecture, formulated in 2008 and remaining unproven to this day, concerns situations where quantum computers can significantly outperform classical machines in speed. According to PSI, this discovery represents the answer to one of the major questions in quantum computing.

AI Enters Mathematical Proofs and Disrupts Career Paths

This announcement fits into a trend where AI contributes to advancements in mathematics. In September, OpenAI declared that it had solved a 90-year-old problem regarding the Navier-Stokes equations, described by the company as "one of the deepest problems at the frontier of mathematics." Terence Tao, a professor at UCLA, stated last month that the "romantic era" of mathematics is over, while noting that new possibilities are emerging, but some things are being lost. According to researchers at MIT, 76% of surveyed master's students and postdoctoral researchers say they are unlikely to remain in academia after graduation, and 76% indicate that advancements in AI are diminishing their motivation for research. More broadly, researchers are questioning the future of the discipline, and some students are reconsidering their career paths.