Artificial intelligence (AI) has potential to accelerate innovation in the energy sector. Exciting examples of this are already evident although uncertainties remain, says Simon Bennett, Energy Technology Analyst at the International Energy Agency (IEA), and Thomas Spencer, Power Sector Modeller at the IEA.
Impressive technological advancements have helped drive down the cost of key energy technologies in recent years. However, to achieve global energy security and emissions goals, existing clean energy technologies need to keep improving and novel energy technologies must reach the market.
IEA’s new workstream on energy and AI involves analysing how the adoption of AI will affect electricity consumption by data centres and how AI can be applied to complex parts of energy systems such as electricity networks.
For energy analysts, a fundamental question is whether AI will cause the rate of technology progress to deviate from current projections. Some see AI as a means to keep current learning rate projections on track. Others see it as a more disruptive force that could make today’s rates look very conservative. To inform this debate, it is necessary to take a closer look at the specific ways AI could boost the pace of innovation.
AI discoveries for energy-related materials
AI has potential to find higher-performing material for a task, which has typically relied on human ingenuity and knowledge of how different compounds behave.
For example, in July 2024, researchers at a US government laboratory published results of a study that used AI to assess 32,5 million possible new solid-state electrolytes for lithium-based batteries and found 23 new versions with the right characteristics.
Scientists in Sweden also used AI to screen 45 million potential new battery cathode molecules and found nearly 4 600 promising options. Researchers using AI have also made breakthroughs in non-battery-related areas, finding they can engineer enzymes for biofuel synthesis, predict high-yielding biofuel feedstocks, identify industry-beating catalysts for hydrogen-producing electrolysers and generate materials for carbon dioxide capture.
Major obstacles remain
There are still challenges to overcome before AI techniques can fulfil their potential for energy innovation. One key issue is data availability. Datasets used today have incomplete information about possible materials, representing a restricted subset of molecules or reactions.
Massive, structured, specialised datasets need to be developed if real-world problems are to be solves. While “synthetic data” can be used to train models to overcome some data gaps, there is no replacement for experimental data and the fastest way to create these datasets is co-operation between laboratories – also at international level.
Another challenge is finding ways for AI to optimise results for more than a simple, narrow set of characteristics and incorporate information that is essential for material to be made into a functional product. Today, substantial human checking and testing is still required.
AI can currently only solve half the challenge by identifying new material for energy application and prototyping, commercialisation and manufacturing can take years – even decades. However, AI could possibly compress these timetables.
The US Department of Energy’s “A-lab”, for example, uses robots that can synthesize energy storage chemicals and offer major performance improvements. This lab can process up to 100 times more samples than a human-run equivalent.
The full potential of AI will only be realised when governments focus on some key emerging issues upfront such as fostering collaboration between laboratories across international borders, supporting commercialisation and investments in skills and equipment.
