Clean energy breakthrough as AI controls fusion reactor

An AI system owned by Google has just "learned" how to control a magnetic field in a fusion reactor – and it could pave the way to generating limitless clean energy.

The breakthrough – where a system designed by DeepMind was able to control the magnetic field in a Swiss tokamak reactor – could pave the way for new designs of fusion reactor, researchers believe.

Tokamak reactors hope to generate fusion energy from a plasma trapped in a magnetic field, but controlling the magnetic field is complex.

The reactors use a powerful magnetic field to confine plasma at extremely high temperatures – hundreds of millions of degrees Celsius, even hotter than the sun’s core – so that nuclear fusion can occur between hydrogen atoms.

Every nuclear reactor currently operating on Earth is a fission reactor – using energy released when heavy atoms such as uranium decay into smaller atoms, a process similar to the one used in the first nuclear weapons.

Read more: Melting snow in Himalayas drives growth of green sea slime visible from space

A fusion reactor works in the opposite way, harvesting the energy released when two smaller atoms join together, releasing tiny, fast-moving particles smaller than atoms.

But to do so, companies need to find a way to harvest energy from a plasma held at millions of degrees Celsius – something that has defied researchers for decades.

Tokamaks form and maintain plasmas through a series of magnetic coils whose settings, especially voltage, must be controlled carefully.

DeepMind's experts developed an AI algorithm that can create and maintain specific plasma configurations and trained it on the simulator at the Swiss Plasma Centre (SPC) of the Ecole Polytechnique Federale de Lausanne (EPFL).

After being trained, the AI-based system was able to create and maintain a wide range of plasma shapes and advanced configurations, including one where two separate plasmas are maintained simultaneously in the vessel.

Read more: A 1988 warning about climate change was mostly right

Finally, the research team tested its new system directly on the tokamak to see how it would perform under real-world conditions.

The DeepMind team wrote in a blog post: "Similar to progress we've seen when applying AI to other scientific domains, our successful demonstration of tokamak control shows the power of AI to accelerate and assist fusion science, and we expect increasing sophistication in the use of AI going forward.

"This capability of autonomously creating controllers could be used to design new kinds of tokamaks while simultaneously designing their controllers. Our work also points to a bright future for reinforcement learning in the control of complex machines."

“DeepMind was immediately interested in the prospect of testing their AI technology in a field such as nuclear fusion, and especially on a real-world system like a tokamak,” said Federico Felici, an SPC scientist and co-author of the study.

Read more: Why economists worry that reversing climate change is hopeless

"We agreed to the idea right away, because we saw the huge potential for innovation," said Ambrogio Fasoli, the director of the SPC and a co-author of the study.

"All the DeepMind scientists we worked with were highly enthusiastic and knew a lot about implementing AI in control systems."

Felici said he was impressed with the amazing things DeepMind could do in a short time when it focused its efforts on a given project.

Brendan Tracey, a senior research engineer at DeepMind and co-author of the study, said: "The collaboration with the SPC pushes us to improve our reinforcement learning algorithms, and as a result can accelerate research on fusing plasmas."

Watch: UK is racing to build first commercial fusion power plant