An unreleased Anthropic model played a central role in settling a quantum computing question that had stood open since 2008, according to Physical Superintelligence Inc. The research lab said Friday that its system, built around the Anthropic model, proved the Aaronson-Ambainis conjecture — a statement about when quantum computers can achieve significant speedups over classical machines.
The conjecture had functioned as an educated guess without a proof for more than a decade. PSI said its research system reached a proof within three days, describing the outcome as the resolution of one of quantum computing’s foundational problems. “Quantum computing still has plenty of mysteries, but now there is one fewer,” the lab said in its announcement.
What the proof shows
According to PSI, the proof demonstrates that certain exponential advantages offered by quantum computing depend on special patterns or structures. That means quantum computers face limits — a possibility that had long been suspected, though the machines would still represent a major upgrade for some tasks. Quantum computers exist today but remain in early stages of development.
The technical work was carried out by David Turturean, a member of PSI’s technical staff, who developed the research system that incorporated the unreleased Anthropic model. PSI did not specify which Anthropic model was used, and Anthropic did not respond to a request for comment.
Reactions from researchers
Researchers who had previously worked on the problem called the proof a “huge result” that resolved a decades-old question. Guy Blanc, Jordan Docter, Carmen Strassle, and Li-Yang Tan of Stanford University said the proof brought everything together in a rather magical way, according to PSI’s announcement.
AI's growing footprint in mathematics
The quantum computing proof is the latest in a series of mathematical breakthroughs attributed to AI systems. In September, OpenAI said it had solved the Navier-Stokes equations, a 90-year-old problem it described as one of the deepest at the frontier of mathematics.
Such advances have prompted reflection among mathematicians about the field’s future. Terence Tao, a prominent professor at UCLA, said last month that mathematics’ “romantic era” had ended, adding that while exciting new things are now possible, some things are being lost.
Math students are also weighing the impact. MIT researchers who surveyed graduate students and postdocs found that 76% reported they were unlikely to remain in academic mathematics after graduating, and the same proportion said AI developments had made them less motivated to pursue research.
For PSI, the announcement marks a notable demonstration of how an unreleased AI model can be harnessed for specialized scientific work. The lab’s system did not merely assist with calculations; it helped produce a formal proof to a conjecture that had resisted resolution for years. The result adds to the broader conversation about AI’s role in mathematical discovery and what that means for the people who study the field.
