David Tortorian, a researcher at Physical Superintelligence (PSI), has managed to solve a scientific dilemma that has eluded scientists for about 18 years, the "Aaronson-Ambainis conjecture," using early access to the Fable 5.2 model from Anthropic. This dilemma relates to the ability of quantum computers to solve problems faster than classical computers.
What actually changed?
This result is a "negative result" in the context of scientific research, as it closed a path that was previously believed to contribute to accelerating quantum computing, thereby directing researchers' efforts toward other more feasible paths. The solution was verified via the mathematical programming language Lean and was also reviewed by a team of senior researchers, including Scott Aaronson, one of the original proposers of the conjecture, who confirmed the validity of the solution despite its complexity.
Why does this news matter?
This achievement represents a shift in how scientific research is conducted, as PSI seeks to build a business model based on "neolabs" (new AI labs). Instead of relying entirely on traditional human methodology, the company uses specialized systems and simulation tools to deepen scientific research. Matt Bainas, CEO of PSI, believes the future is heading toward making theoretical physics a field primarily dominated by machines.
Challenges and Future Prospects
Despite success in fields such as mathematics and computing, the application of AI in physics and biological sciences still faces greater challenges, as nature requires experimental verification that is not limited to computer code. PSI is currently working on developing workflows that integrate computational results with effective experimental verification. These developments also raise questions about new ethical and academic standards in light of the accelerating pace of AI-supported scientific discoveries, as academic circles, such as MIT, have begun holding summits to discuss how to deal with this new reality.