AI-Assisted Research Cracks a Fluid Mechanics Problem
On August 30, a report revealed that scientists at the University of Colorado Boulder turned to Anthropic's Claude AI to solve a fluid mechanics challenge that had resisted their laboratory for a year and a half. The breakthrough took five weeks, and the process was punctuated by repeated AI mistakes.
The study, authored by associate professor Ankur Gupta and graduate student Arkava Ganguli, appeared in the Journal of Fluid Mechanics. It centers on the electrophoresis of nanoparticles—the motion of charged particles under an electric field—and shows that changing a particle's shape, such as stretching a circle into an oval, alters its velocity in that field. The team also determined that tiny surface features on a nanoparticle have no effect on its speed.
This finding indicates that for nanoparticles, roughly one-thousandth the width of a human hair, the long-standing rule established more than a century ago by physicist Marian Smoluchowski applies differently. Such insights are part of a growing trend: researchers are increasingly relying on large language models to handle computation, code generation, and visualization, even while acknowledging that AI outputs require careful scrutiny.
What the Findings Mean
Claude's role involved performing complex mathematical calculations, writing code, and producing graphic illustrations. On occasion, however, the AI fabricated convincing errors that had no basis in the original data. The project demonstrates how modern tools can contribute to difficult scientific problems, despite their imperfections.
This effort underscores the growing importance of artificial intelligence in research, showing how advanced systems can help solve complex challenges while still needing human oversight.
The paper's DOI is 10.1017/jfm.2026.11948. Ultimately, the work showcases the practical benefits of integrating AI into the scientific process, especially in fluid mechanics. Even when the technology stumbles, its capacity for computation and data analysis opens new possibilities for researchers and may accelerate discoveries across related fields.
This development in fluid mechanics is part of a broader trend where AI tools are increasingly utilized in complex scientific inquiries. For instance, the recent advancements made by Anthropic's unreleased model in addressing the Riemann Hypothesis further illustrate the potential of AI in tackling long-standing mathematical challenges. To learn more about these exciting breakthroughs, visit how AI is reshaping mathematical research.