A recent MIT graduate working at the startup lab Physical Superintelligence, known as PSI, has used early access to Anthropic’s Fable 5.2 model to solve a long-standing question in quantum computing. David Turturean, who deferred plans to pursue a master’s degree at MIT’s computer science and AI lab to join PSI, tackled the Aaronson-Ambainis conjecture, a central problem in quantum query complexity that has persisted for nearly two decades.
The Aaronson-Ambainis conjecture concerns the ability of quantum computers to solve certain problems faster than classical computers. Scott Aaronson, one of the researchers who originally posed the problem, stated that while he does not fully understand every step of the proof, he believes it is correct because it was autoformalized in Lean, a coding language for mathematics, and vetted by a team of top researchers with whom PSI shared the solution. The negative result closes off one specific avenue that quantum computers might have pursued to achieve high-speed computation, though other pathways remain open.
Key takeaways
- PSI researcher David Turturean used Anthropic’s unreleased Fable 5.2 model to solve the 18-year-old Aaronson-Ambainis conjecture.
- The proof was autoformalized in the Lean coding language and reviewed by top researchers including Scott Aaronson.
- PSI currently generates revenue by using its tools to optimize data center variables such as heat, power, water, and sound.
- The startup ultimately aims to convert theoretical physics breakthroughs into tangible technologies through experimental validation loops.
Building a business on AI-driven science
The breakthrough highlights the ongoing shift in the scientific community driven by advanced artificial intelligence problem-solving capabilities. PSI CEO Matt Pines noted that an effective research system requires robust harnesses, tools, memory systems, and simulation infrastructure to approach complex problems with both breadth and depth. While the startup's current revenue stems from optimizing data center performance where physical variables directly impact community and site metrics, the long-term vision involves translating theoretical physics breakthroughs into real-world technologies.
Transitioning computational results into the physical world presents unique challenges compared to pure mathematics or software. According to Pines, physics differs from math because it interacts directly with the real world, meaning that nature ultimately tests models of reality. PSI is focusing on how to efficiently take computational results and validate them through physical experiments, turning successful ideas into functional devices.
Emerging standards and the future of research
As neolabs for science continue to emerge as pathways for potentially groundbreaking discoveries, researchers and institutions are still establishing best practices. Because PSI held onto its result until Anthropic's model was nearly ready for public release, the team was able to consult relevant experts and maintain academic transparency. Jim Halverson, a Northeastern University professor who advises PSI, noted that academic standards for this new pace of science are still evolving, prompting upcoming discussions among leading physicists.
For Turturean, the experience underscored the immense weight and capability of modern AI systems in handling complex intellectual tasks. As artificial intelligence continues to accelerate scientific inquiry, researchers are actively working to balance unprecedented computational speed with traditional expectations of academic explanation and peer review.
Comments