PSI Solves 18-Year-Old Aaronson-Ambainis Conjecture with Fable 5.2

PSI Solves 18-Year-Old Quantum Conjecture with Fable 5.2 and Establishes a Scientific Business Model

Startup Physical Superintelligence (PSI) announced that its technical team member David Turturean, a recent MIT graduate, used early access to Anthropic Fable 5.2 to solve the Aaronson-Ambainis conjecture, open for 18 years in quantum query complexity. The result was published after academic verification and automated formalization in Lean, signaling a fundamental shift in scientific discovery methodology.

Announcement Details and Technical Architecture

PSI relies on what CEO Matt Pines calls an integrated research system including 'harnesses', tools, memory, and simulation architecture, aiming to attack problems with depth and breadth and test multiple alternatives. The solution to the Aaronson-Ambainis conjecture came through this system and was automatically verified in Lean, prompting Scott Aaronson to say he does not fully understand the proof but trusts its correctness. The result is negative and closes one path for quantum computing without closing all paths.

Currently, PSI funds itself through data center optimization, simulating variables of heat, energy, water, and sound because site performance and community impact are a direct function of its physics. The long-term goal is to convert theoretical physics breakthroughs into actual technologies through loops: AI proposal, experimental verification, then device construction. Pines emphasizes that 'nature has a voice' in physics models, making experimental verification a necessity, not a luxury.

Benchmark and Competitive Performance

PSI's solution stands out amid a wave of automated proofs announced by major labs, as Aaronson noted surprise that the conjecture was absent from OpenAI's 'Mathocalypse dump' weeks earlier. This places PSI in direct competition with major AI labs on solving open mathematical problems. Meanwhile, Discovered Materials launched a $1,000,000 challenge for applied physics, reflecting market awareness that physics is harder than mathematics due to the need for experimental verification. Harbinger also raised $240,000,000 after expanding into defense, indicating investor appetite for deep tech.

Industry Impact and Enterprise Adoption

PSI does not currently offer a product aimed at Arab developers, but its model opens indirect opportunities: physics-based data center optimization tools could contribute to energy efficiency for regional data centers in the Gulf and Egypt, where cooling and energy costs are a major axis. Relying on 'harnesses' and external tools instead of a single massive model reduces token costs and allows building local research systems with limited resources. For Arab startups, the lesson is that specializing in a narrow scientific problem with rigorous verification can create defensible value.

The 'neolabs for science' model that PSI follows—startup labs using specialized AI models with tools, memory, and simulation architecture—is still evolving. PSI works with academic researchers like Jim Halverson of Northeastern University to set transparency and publication standards, but best practices are still being shaped. The broader implication is that AI-driven discovery is moving from demos to business models, with verification and experimental loops as the core moat.

Conclusion

PSI's solution of the Aaronson-Ambainis conjecture with Anthropic Fable 5.2 is a milestone in AI-assisted science: an 18-year-old open problem solved, verified in Lean, and reviewed by leading researchers. The company's business model—data center optimization funding deeper physics research—and the $1,000,000 Discovered Materials challenge show that the market is beginning to reward verified scientific progress. For enterprises and developers, the takeaway is clear: the next wave of AI value lies in specialized systems that combine models with tools, memory, and rigorous experimental validation.

Media Source: TechCrunch AI | Fact Verification & Analysis: AI Tools Oasis