Researchers at Princeton University and the U.S. Department of Energy’s Princeton Plasma Physics Laboratory have developed an AI framework that can monitor and control fusion plasma in milliseconds. Called PACMAN, the system can process live plasma measurements and adjust fusion conditions far faster than a human operator. In experiments on the DIII-D tokamak, PACMAN successfully controlled heating systems, adjusted plasma density and rotation, and detected potentially dangerous instabilities. In one experiment, the AI predicted a tearing-mode instability about 200 milliseconds before it appeared, allowing researchers to change the plasma and prevent the instability from developing. The system could help make fusion experiments faster, safer, and more efficient by responding to rapid plasma changes that humans cannot react to quickly enough. Its modular design could also allow AI models to be added or replaced more easily across different fusion machines. However, humans remain responsible for setting the system’s objectives and safety limits.
