Emory University AI Discovers Novel Non-Reciprocal Force Laws in Plasma Physics

On April 23, 2026, a team of physicists at Emory University published a paper in PNAS announcing that they used physics-inspired custom neural networks to discover previously unidentified non-reciprocal force laws from three-dimensional motion trajectories of dusty plasma particles, with the model's prediction accuracy exceeding 99 percent.

This breakthrough marks AI's evolution from a data analysis tool to a discoverer of natural laws, elevating it from a passive data processor to an active knowledge creator.

Understanding non-reciprocal forces: Plasma is the fourth state of matter (beyond solid, liquid, and gas), containing free electrons and ions. More than 99 percent of visible matter in the universe is plasma. Dusty plasma specifically refers to plasma containing charged dust particles, and non-reciprocal forces mean that the forces between particles are asymmetric.

This research breakthrough demonstrates that AI can not only simulate physical systems but also independently discover new physical laws that humans have not yet recognized. This capability opens new pathways for scientific discovery.

The implications for embodied AI are significant: as robots and AI systems interact with the physical world, the ability to discover and understand fundamental physical laws becomes crucial for developing more capable and adaptable systems.