A long-standing physics puzzle is finally solved thanks to a collaboration between humans and artificial intelligence. Nobel laureate Giorgio Parisi and physicist Francesco Zamponi spent a decade searching for a formal proof regarding the jamming transition. This phenomenon explains how fluid systems, like spheres in a box, suddenly lock into a rigid state.
After years of manual research, the pair turned to the large language model Claude. They prompted the AI to replicate numerical calculations from a 2014 paper. Once the model demonstrated its capability, the team asked it to explain the relationship between two specific parameters where the sum equals one. While the AI required several iterative refinements, it provided the conceptual breakthrough that had eluded the researchers for years.
This outcome demonstrates how machine learning models serve as a tool for high-level scientific reasoning. Zamponi noted that the AI offered a simpler solution than the complex symmetries the team had anticipated. The researchers published the full transcript of their interaction with the model alongside their findings, setting a standard for transparency in AI-assisted research.
Beyond this specific physics application, the event highlights a shift in how research is conducted. Rather than replacing human intuition, the technology acts as a telescope for the mind. This approach helps scientists navigate complex literature and test hypotheses at unprecedented speeds, though it also raises questions about maintaining quality control in peer-reviewed journals.