Researchers are increasingly employing artificial intelligence to help them solve some of the biggest challenges facing the energy sector – including, ironically enough, the massive spike in energy demand caused by large language models themselves. The current and projected rise in energy demand from AI data centers is driving a wave of investment into next-gen energy alternatives that can create huge amounts of baseload power without emitting large amounts of greenhouse gases. One of these potential ‘silver bullet’ technologies is nuclear fusion, which has been gaining major ground in laboratories in recent years, thanks in part to AI tools.

To this end, scientists at the Ames National Laboratory in Ames, Iowa are building an AI tool dedicated to modeling how different materials will work in nuclear fusion processes in order to refine research approaches and make the scientific process as well as fusion systems more efficient. The tool, which is called DuctGPT, was adapted from a previous National Institute of Standards and Technology model called AtomGPT. The Duct version uses large language modelling in combination with physics modeling to find materials suitable for the harsh environment of a nuclear fusion reactor.

Nuclear fusion, the process which powers our own sun, requires ultra-high temperatures that most materials have no chance of withstanding. In addition to being able to stand up to temperatures measuring in the thousands, millions, or even hundreds of millions of degrees, these materials also need to remain ductile enough for manufacturing purposes. Finding the right material presents a major hurdle for achieving commercial nuclear fusion, and a major opportunity for the scientific team that is able to crack the code, releasing what could be virtually limitless clean energy. Finding the right materials will require the exploration and modelling of a huge range of possible alloy compositions.

This is exactly the type of project that large language models are suited to. “Finding new materials, catalysts or processes that can produce stuff more efficiently is the sort of ”needle in a haystack’ problem that AI is ideally suited to,” the Financial Times reported last year in an article musing about “How AI might save more energy than it soaks up.”.

The new AI tool is already showing enormous promise for the burgeoning field of nuclear fusion research. The team behind DuctGPT reports that the time needed to discover new alloys for nuclear fusion experiments has already been slashed from months of research down to mere hours. “Now when you ask it, ‘I want to design a material for fusion that has all x, y, z properties that are critical for use in fusion reactors,” Tell me the combination of elements which satisfy the criteria,’ it will give you those combinations of elements with properties,” said Ames Lab Scientist Prashant Singh.

While DuctGPT represents one of the latest and most promising applications of large language models in the field of AI research, it’s certainly not the only one. Another new AI tool called Diag2Diag is being used to help monitor and control plasma in fusion experiments, specifically in order to avoid what is known as the Edge Localized Mode (ELM). ELM is a condition of instability that rapidly breaks down the materials around the plasma, causing major issues for huge and costly plasma experiments like Europe’s ITER and China’s EAST.

Meanwhile, in the United Kingdom, the government is investing £45 million (approx. USD $60 million), to build an AI supercomputer at the UK Atomic Energy Authority’s Culham campus in Oxfordshire. The machine – named Sunrise – is expected to begin operations next month. “Officials say the machine will help scientists better understand the complex physics at work in fusion reactors,” states an Interesting Engineering report from March. “By combining advanced computing with artificial intelligence models, the supercomputer could allow researchers to test ideas virtually before building costly experimental systems.”

Together, these tools could majorly advance the field of nuclear fusion research, and not a moment too soon. Investing in unproven technology may be a risky bet, but nuclear fusion is closer to reality than ever before, and breakthroughs are piling up as the field becomes more competitive and Big Tech throws its weight behind the issue. The energy monster that AI is creating is so massive and unprecedented that the tools we use to fix it must also be unprecedented – hence why AI solutions might be the only way to fix AI problems.

By Haley Zaremba for Oilprice.com

https://oilprice.com/Energy/Energy-General/AI-Is-Being-Used-to-Fix-Its-Own-Energy-Problem.amp.html