DeepMind AI and Nuclear Fusion?
Among Google DeepMind's weirdest of headlines so far in 2022
At AiSupremacy I’m covering the intersection of Artificial Intelligence and breaking news.
This week DeepMind AI created yet another unusual headline. It appears that DeepMind has trained an AI to control nuclear fusion. So let’s get into it.
Accelerating fusion science through learned plasma control
More specifically DeepMind’s AI can control superheated plasma inside a fusion reactor. You can read DeepMind’s blog about it here.
The Google AI lab has been thinking about solving the energy crisis. To solve the global energy crisis, researchers have long sought a source of clean, limitless energy. Nuclear fusion, the reaction that powers the stars of the universe, is one contender.
I like the scientific implications of the kind of problems A.I. is working on related to all of humanity or a AI for Good broad based perspective.
The London-based AI lab, which is owned by Alphabet, announced Wednesday, February 16th, 2022, that it has trained an AI system to control and sculpt a superheated plasma inside a nuclear fusion reactor.
The company says it collaborated with the Swiss Plasma Center at the EPFL technical university in Switzerland "to develop the first deep reinforcement learning (RL) system" devoted to the tools researchers are using to assess the nuclear fusion's viability as an energy source.
The Potential of Nuclear Fusion Harnessed by A.I.
Nuclear fusion, a process that powers the stars of the universe, involves smashing and fusing hydrogen, which is a common element of seawater.
DeepMind claimed that the breakthrough, published in the journal Nature, could open new avenues that advance nuclear fusion research.
The DeepMind team has made news for years by using increasingly sophisticated AI to master iconic games like chess, shogi, and Go, as well as to solve “real world” problems.
It’s just nice to see them actually thinking a bit more about humanity’s challenges ahead finally. The London-based workgroup was formed in 2010 and purchased by California-based Google in 2014. They were bleeding cash by have worked on becoming more profitable (mostly due to interactions with their parent company).
The science behind it is actually pretty interesting. That reinforcement learning system was designed to "autonomously discover how to control" a tokamak, which DeepMind says is "a doughnut-shaped vacuum surrounded by magnetic coils" that is "used to contain a plasma of hydrogen that is hotter than the core of the Sun."
Do you enjoy A.I. articles at the intersection of breaking news, than help me continue to write on the subject.
British artificial intelligence scientist and DeepMind CEO Demis Hassabis.
In essence on Nuclear Fusion, it’s a process, which releases vast amounts of energy, has been touted as a potentially limitless source of clean energy, but a number of technical challenges still need to be overcome.
The magnets in these tokamaks are used to “contain” the volatile hydrogen plasma, which is hotter than the core of the sun. Controlling the magnetic coils currently requires multiple layers of complex control systems.
In new research published in the peer-reviewed journal Nature, the DeepMind team explains how they used deep reinforcement learning—a subfield of machine learning where a system can learn from its own decisions—to help to magnetically control the hot, usually hydrogen, plasma in a tokamak reactor.
Further reading: https://www.nature.com/articles/s41586-021-04301-9
However, the plasmas in these machines are inherently unstable, making sustaining the process required for nuclear fusion a complex challenge.
For example, a control system needs to coordinate the tokamak's many magnetic coils and adjust the voltage on them thousands of times per second to ensure the plasma never touches the walls of the vessel, which would result in heat loss and possibly damage.
DeepMind has Doubled Down to Augment Science in the 2020s
A.I. can certainly be helpful in engineering and science to help solve these fundamental problems for the future smart cities. As we’ve noted in this Newsletter A.I. is increasingly being used in everything from climate change to physics.
DeepMind said it has developed a reinforcement learning AI system that can control the magnets and change their voltage thousands of times per second. This is an important demonstration of A.I.’s ability to help control scenarios that are potentially dangerous.
DeepMind’s unnamed AI, developed on a virtual simulator, has been used around 100 times on a tokamak at the Swiss Plasma Center known as the Variable Configuration Tokamak. It controlled the magnets in the tokamak for two seconds, which is the maximum amount of time the reactor can run before it overheats. It was quite a large undertaking, as roughly 10-20 people from DeepMind worked on the AI system together with around 5-10 people from EPFL.
“In the last two years DeepMind has demonstrated AI’s potential to accelerate scientific progress and unlock entirely new avenues of research across biology, chemistry, mathematics and now physics.”
FOUNDER & CEO, DEEPMIND
We can imagine the scientific breakthroughs that A.I. will enable in the future. Harnessing energy process that powers stars will be as impactful on human history ‘as the adoption of electricity’
Nuclear fusion is the process by which our sun and other stars power themselves, however after decades of research it remains frustratingly out of reach.
DeepMind set out to crack artificial general intelligence, which is often referred to as the holy grail of AI, just like Microsoft backed OpenAI. This scientific example however shows how reinforcement learning (RL) can truly accomplish important feats for civilization at large.
Nuclear fusion occurs when two atoms smash together to form a heavier nucleus, a process that releases a massive amount of energy in the form of plasma. Inside stars, this plasma is held together by gravity. Here on Earth, scientists must rely on powerful lasers and magnets, such as one being developed by MIT and the Bill Gates-backed Commonwealth Fusion Systems.
DeepMind claims this work is another powerful example of how machine learning and expert communities can come together to tackle grand challenges and accelerate scientific discovery. It’s really pretty good PR for the team.
Basically, the DeepMind team has taken the reins on the magnetic coils themselves and used deep machine learning to dynamically adjust each electromagnet’s encoded behaviors.
What Will AI Labs Actually Ever Accomplish?
While Google has found uses for DeepMind’s AI, its technology has not been widely applied elsewhere. It’s still an extremely expensive Moon shot AI lab for Alphabet. It does some good research but continues to struggle to find real-world applications. I am not optimistic that it will ever approach anything remotely like AGI.
DeepMind CEO Demis Hassabis said in a statement that the company has demonstrated AI’s potential to accelerate scientific progress and open new avenues of research across biology, chemistry, mathematics and now physics. Certainly for A.I. at the intersection of science this is super interesting.
It’s a huge investment for Alphabet, one of the companies that continues to attract the most talented people in A.I., machine learning and deep learning.
But at what cost? DeepMind employs about 1,000 people worldwide, including some of the world’s leading AI research scientists, who can command annual salaries of more than $1 million.
DeepMind like OpenAI are somewhat controversial organizations typically with very white male leadership. DeepMind inherits to some degree Google’s own historically sexist and racist culture.
These top people, who often have Ph.D.s from the likes of Oxford, Cambridge, Stanford and MIT, can command this sort of money because they’re also sought after by Big Tech companies like Facebook, Apple, Amazon and Microsoft.
Still above the politics and lack of inclusion in the world’s top AI firms, the potential impact of this work is substantial. The breakthrough, published in the journal Nature, could help physicists better understand how fusion works, and potentially speed up the arrival of an unlimited source of clean energy.
In a future article I’m going to rank the major A.I. labs of the world with the top A.I. Colleges and Universities.
I cannot continue to write without tips, patronage and community support from you, my readers and audience. I want to keep my articles free for the majority of my readers.
So by subscribing you are essentially helping fund a network of Newsletters whose aim is to inspire and inform.
AiSupremacy is the fastest Substack Newsletter in AI at the intersection of breaking news. It’s ranked #1 in Machine Learning as of January 22nd, 2022.
Thanks for reading!