AI Shatters 18-Year Quantum Computing Barrier: A New Era for Scientific Discovery

In a development that signals a profound shift in the landscape of theoretical physics, David Turturean, a recent graduate of the Massachusetts Institute of Technology (MIT) and a key member of the technical staff at the startup laboratory Physical Superintelligence (PSI), has successfully solved an 18-year-old open problem in quantum computing. Utilizing early access to Anthropic’s advanced Fable 5.2 model, Turturean provided a solution to the Aaronson-Ambainis conjecture—a foundational puzzle concerning the computational advantages of quantum systems over their classical counterparts.

The achievement is more than a mere technical milestone; it is a harbinger of a future where artificial intelligence functions not just as a tool for data processing, but as an engine for original scientific discovery. For Turturean, who deferred his plans for a master’s degree at MIT’s prestigious Computer Science and Artificial Intelligence Laboratory (CSAIL) to join the nascent team at PSI, the experience was transformative.

“I’m totally accepting the idea that this might be the last paper I ever put out,” Turturean remarked during an interview with TechCrunch. His sentiment reflects a growing realization within the scientific community: as AI models gain the ability to navigate complex logical and mathematical landscapes, the traditional paradigm of the human researcher as the sole architect of new theories is being challenged. “In a few months, like, why would you even put out a paper?” he mused, highlighting the rapid acceleration of AI-driven research capabilities.

The problem itself, the Aaronson-Ambainis conjecture, has long been a subject of intense scrutiny in the field of quantum query complexity. It concerns the limitations of quantum computers, specifically addressing the degree to which they can outperform classical algorithms on certain types of problems. Scott Aaronson, one of the original researchers who proposed the conjecture nearly two decades ago, expressed a complex mix of emotions regarding the breakthrough.

Writing on his personal blog, Aaronson admitted that while he did not fully grasp the intricacies of the proof provided by the AI, he was confident in its validity. His confidence was bolstered by the fact that the solution was autoformalized in Lean—a sophisticated programming language used for mathematical verification—and subsequently vetted by a team of top-tier researchers with whom PSI shared the findings.

“A proof of the full Aaronson-Ambainis conjecture was one of the last remaining open problems in quantum query complexity that everyone seemed to agree was centrally important,” Aaronson wrote. He described his reaction as a confluence of satisfaction, awe, disappointment, and fatigue. He also noted his surprise that the problem had not been solved during the recent wave of AI-driven mathematical breakthroughs, sometimes referred to as the “Mathocalypse.” “Surely they tried it,” Aaronson wrote, questioning why this specific problem had remained elusive until now. “So was it really so much harder than all the problems that were solved?”

The answer to that question appears to lie in the methodology employed by PSI. The startup is distinguished by its focus on creating specialized model harnesses and scaffolding, designed specifically to tackle high-level scientific problems. Rather than relying on a general-purpose AI, PSI builds infrastructure that allows for a more rigorous, structured, and deep-dive approach to problem-solving.

Matt Pines, CEO of Physical Superintelligence, explains that the company’s philosophy is rooted in building an ecosystem of research. “An effective kind of research system should have lots of things related to harnesses, lots related to tools, memory systems, simulation infrastructure,” Pines noted. “You want to be able to really go after a problem at breadth and depth, and be able to consider a wide range of potential alternatives.”

While the theoretical breakthrough regarding the Aaronson-Ambainis conjecture is a significant victory for the company’s intellectual capabilities, PSI’s current operational model is grounded in more immediate, practical applications. The firm generates revenue by applying its computational tools to the optimization of data centers. By simulating the complex variables of heat, power, water, and sound, PSI works to refine the efficiency of physical infrastructure. As Pines describes it, “the performance of that site as well as its impact on the community are a direct function of its physics.”

However, the ultimate objective for PSI remains the translation of abstract theoretical physics into tangible, real-world technology. Reflecting on his recent interactions with potential investors, Pines noted that the skepticism surrounding the role of AI in theoretical science is rapidly evaporating. “Six to nine months ago, when I was talking to investors, I had to sort of stretch to validate that conditional,” he explained. “Like, ‘no, no, no, trust me, pretty soon, theoretical physics is going to be mostly the domain of machines.’ And it’s not there yet, but it’s on its way.”

Despite the excitement surrounding AI’s potential, there is a clear distinction between the realms of pure mathematics and the physical sciences. While AI has demonstrated an extraordinary ability to claim victories in fields like computer science and math—where results can be expressed and validated through code—applying these same methods to the life sciences or experimental physics presents a more daunting challenge. Because these fields are subject to the unpredictable nature of the physical world, computational results must be validated through rigorous, real-world experiments.

The industry is already responding to these challenges. Just last week, the startup Discovered Materials announced a $1 million grand challenge, encouraging researchers to leverage AI to unlock solutions for complex problems such as superconductivity and heat dissipation.

Pines emphasizes that the gap between simulation and reality is the next great frontier for the industry. “At the end of the day, physics is different than math because it’s the real world… nature gets a vote in your models of reality,” he said. The strategy at PSI is to master the feedback loop: taking computational results, verifying them, and then translating them into physical applications. “If it’s a true fact of nature, that’s a trick that you can exploit,” Pines continued. “You could build a device out of it, right? The AI suggests an idea; it allows you to do something in the world you weren’t able to do before… and so each of those are kind of loops that PSI is building.”

As these "neolabs for science" gain prominence, the community is grappling with the ethical and procedural implications of this new age of discovery. Because PSI had the foresight to hold its result until Anthropic’s Fable 5.2 model was nearing public release, the team had the necessary time to engage with experts and ensure the integrity of the proof.

Jim Halverson, a professor at Northeastern University and an advisor to PSI, sees this collaborative approach as essential for the future of scientific integrity. He is scheduled to attend a summit at MIT next week, where leading physicists will gather to discuss how the discipline should adapt to the unprecedented pace of AI-driven research. “As an academic, we wanted to play our role in doing things the right way,” Halverson said. “Best practices are still emerging… We figured that transparency is a good way to go, and contacting experts is a good way to go.”

For Turturean, the experience of shepherding an AI-generated solution through the gauntlet of academic scrutiny was a sobering reminder of the human role in this new scientific order. “We had some time to sit on it… and it was the first time that I actually took one of my AI-elicited solutions on print, and I read through it, and I was like, okay, how can I make this… be presented more naturally for a human?”

As he prepares to explain the result to the broader academic community, Turturean remains struck by the magnitude of the shift. The ability to solve an 18-year-old conjecture with the assistance of a machine has not diminished his sense of wonder; rather, it has clarified his vision of the future. “And I think it also made me realize,” he concluded, “oh my God, the weight of what AI can do.”