DeepMind AI Achieves Breakthrough in Nuclear Fusion Plasma Control
DeepMind and the Swiss Plasma Center successfully use deep reinforcement learning to control and sculpt superheated plasma inside a tokamak reactor. This breakthrough opens new avenues for advancing nuclear fusion research.
DeepMind partners with the Swiss Plasma Center at EPFL to develop the first deep reinforcement learning system capable of autonomously controlling nuclear fusion plasma. By utilizing a simulated environment, this AI system learns to coordinate the complex magnetic coils of a tokamak, a doughnut-shaped device that holds hydrogen plasma hotter than the center of the Sun.
The artificial intelligence successfully maintains the superheated plasma in a steady state without touching the reactor walls, preventing heat loss and potential damage. Furthermore, the system demonstrates the ability to accurately sculpt the plasma into various shapes, which gives scientists a powerful new tool to investigate how plasma reacts under different structural conditions.
This achievement marks a significant step forward in the quest for clean, limitless energy by solving a major instability challenge in fusion reactors. By applying advanced AI to physics problems, DeepMind continues to demonstrate how machine learning accelerates scientific progress and unlocks entirely new avenues of research in the global pursuit of practical fusion power.