AI on Track to Solving Unlimited Energy

Artificial intelligence at Princeton University has made significant progress in nuclear fusion technology by effectively predicting and managing plasma instabilities, bringing us one step closer to unlimited clean energy.

Share
AI on Track to Solving Unlimited Energy

Researchers at Princeton University in the US have used artificial intelligence (AI) to overcome one of the biggest challenges in producing nearly unlimited clean energy through nuclear fusion. An innovative AI system has been developed to predict and manage plasma instabilities in real time during fusion reactions, enabling use at a larger scale.

Egemen Kolemen, a physicist at the Princeton Plasma Physics Laboratory, explains that by drawing on data from previous experiments instead of traditional physics-based models, the AI has been able to develop an effective control policy that maintains a stable, high-performance plasma regime in an actual reactor.

Nuclear Fusion One Step Closer to Reality

Nuclear fusion, often seen as the holy grail of clean energy, offers the potential to generate enormous amounts of energy without using fossil fuels or creating hazardous waste. This process mimics the natural reactions that take place in the sun, but has historically been technically difficult to master.

In 2022, researchers at Lawrence Livermore National Laboratory in California achieved a historic milestone when they succeeded, for the first time, in producing more energy from nuclear fusion than was used to start the reaction, even though the amount of energy was comparable to what's needed to boil a kettle.

AI Predicts and Manages Plasma Instabilities

This latest breakthrough means artificial intelligence can now identify instabilities in plasma three hundred milliseconds before they occur, giving enough time to make adjustments and keep the plasma stable.

SanKyeun Kim, a researcher involved in the project, points out that the ability to predict these instabilities improves the ability to control the reactions, compared to previous methods that were more reactive and required quick corrections to prevent disruptions in the plasma.

The research, published in Nature under the title "Avoiding fusion plasma tearing instability with deep reinforcement learning," marks an important advance in the ambition to make nuclear fusion a practical, scalable energy source.

Source: The Independent UK