PSI’s Breakthrough in Quantum Computing
Physical Superintelligence (PSI) has announced a significant achievement by utilizing early access to Anthropic’s Fable 5.2 model to resolve the 18-year-old Aaronson-Ambainis conjecture. This conjecture addresses the potential speed advantages of quantum computers over classical machines.
Details of the Solution
David Turturean, a recent MIT graduate and a member of PSI’s technical staff, played a key role in this development. He noted that the solution emerged from extensive work with the model and its associated toolchain. Turturean, who postponed his plans for a master’s degree at MIT to join PSI, remarked on the weight of this outcome, suggesting it may challenge the necessity for future academic papers in certain contexts.
Scott Aaronson’s Reaction
Scott Aaronson, one of the original proposers of the conjecture, expressed that while he does not fully comprehend the proof, he accepts its validity due to its autoformalization in Lean and the collaborative explanation provided by PSI’s research team. He indicated that this proof effectively closes one avenue for quantum computers to achieve specific high-speed solutions, while still leaving other possibilities open for exploration.
PSI’s Business Model and Future Plans
PSI CEO Matt Pines described the startup’s approach as integrating specialized harnesses, tools, memory systems, and simulation infrastructure, enabling models to explore problems both broadly and deeply. Currently, PSI generates revenue by optimizing data centers, simulating variables such as heat, power, water, and sound. Pines is also focused on a long-term vision of translating theoretical physics breakthroughs into practical technologies.
Challenges in Applying AI to Physics
Pines acknowledged the challenges of applying AI in less abstract fields like physics and life sciences, where results must ultimately be validated by nature. He referenced a recent initiative by Discovered Materials, which launched a $1 million grand challenge for AI research on complex issues like superconductivity and heat dissipation, highlighting the difficulties faced in these domains.
Peer Review and Emerging Standards
PSI maintained confidentiality regarding their results until the imminent release of Anthropic’s model, sharing the formalized proof with relevant experts. Jim Halverson, a professor at Northeastern University and advisor to PSI, noted that the academic community is still establishing best practices for transparency and expert consultation. Halverson is set to attend a summit at MIT where leading physicists will discuss the implications of the rapid advancements in AI-driven science.
Reflections on AI’s Role in Research
Turturean reflected on the process of reviewing an AI-generated solution, emphasizing the importance of presenting such work in a manner accessible to human readers. He acknowledged the profound implications of AI’s capabilities in research, as PSI continues to balance its data-center optimization business with the pursuit of experimental validation and technology translation pathways.
Original source: TechCrunch AI