Physicists and AI Model Claude 'Collaborate' to Prove 10-Year-Old Jamming Conjecture

A long-standing mathematical enigma in the physics of complex systems, which had eluded resolution for over a decade, has finally been cracked through an extraordinary collaboration. This groundbreaking study involved two theoretical physicists, Nobel laureate Giorgio Parisi and Francesco Zamponi from La Sapienza University of Rome, working alongside the artificial intelligence model, Claude. Published in the Journal of Statistical Mechanics: Theory and Experiment, their findings not only provide a crucial scientific answer but also offer a tangible demonstration of how AI is fundamentally reshaping the landscape of scientific inquiry and research methodologies. The problem at hand revolves around the concept of "jamming" in physics, a phenomenon describing how a system, initially fluid, suddenly becomes rigid while maintaining a disordered state, akin to a traffic jam of particles. While originally applied to materials like foams and granular matter, the concept's surprising generality has led to its adoption in diverse fields, including neuroscience and artificial intelligence itself. In 2014, Parisi, Zamponi, and their collaborators developed a theoretical description of jamming, observing a peculiar relationship: two mathematical parameters, 'a' and 'b', consistently summed to 1 in numerical calculations, a finding that mirrored physical laws derived from an independent theoretical approach by French physicist Matthieu Wyart. Despite the extraordinary accuracy of these numerical results, a formal mathematical proof for the a+b=1 relationship remained elusive for years. Researchers were convinced that a deeper, hidden structure within the theory must underpin this apparent simplicity, yet all attempts to uncover it proved fruitless. The problem, though persistent, gradually faded from the forefront of research for many, but not for Parisi, who remained determined to find its proof. He saw the emergence of generative AI models as a unique opportunity, identifying this old problem as an ideal test case due to its clear conjecture and numerically known answer. Claude was specifically chosen for this task, as Zamponi noted, due to its "somewhat more advanced mathematical reasoning abilities." The initial interaction with Claude wasn't to demand a proof directly. Instead, Parisi first prompted the AI to reproduce the numerical calculations developed by his group over a decade prior, aiming to gauge its capacity for tackling real mathematical challenges. Once Claude successfully replicated the results, the logical next step was to ask: "If a+b equals 1, can you also prove why?" Remarkably, Claude swiftly generated an initial idea that, despite containing some errors requiring subsequent rounds of verification and revision by the human researchers, captured the correct underlying intuition. The most surprising revelation, however, wasn't just the AI's ability to contribute to the proof, but the nature of the solution itself. The physicists had spent years searching for a profound, complex explanation, expecting it to unveil new mathematical structures or unknown symmetries. Instead, the answer proved to be far simpler than anticipated, "right there," as Zamponi described, highlighting a blind spot in their previous human-led investigations. This collaboration ultimately confirms that two distinct theoretical approaches to jamming, developed independently, indeed converge to describe the same fundamental physical laws.