AI can help deliver the cleaner and cheaper energy needed for the future

Cheaper, cleaner, faster, better: that should be the motto of the energy industry for the current times. Adipec, the world’s biggest energy conference, held last week in Abu Dhabi, showcased these needs , and many solutions. The dominant theme of the event, artificial intelligence, can deliver many of those solutions . The main outlines of the future energy system, to deliver on climate goals, have a high level of consensus among scientists, consultancies, international agencies and governments. They include a major boost in energy efficiency, a huge scale-up in renewable energy (mostly solar and wind), batteries to store power, and the conversion of ground transport to electric vehicles. A limited amount of carbon capture, hydrogen and nuclear power are used in the “hard-to-abate” sectors, such as heavy industry, for long-duration reliable electricity supply, and to supply chemical fuels for long-distance shipping and aviation. AI can help to deliver this promised renewable utopia, based on electrons, flowing from solar panels through high-voltage cables to batteries and electric motors. Efficiency is perhaps the most immediate opportunity. A network of intelligent sensors and controls can detect and shut off methane leaks, ensure motors are running at optimal speeds, manage heating and air-conditioning systems. New materials can cool by radiating heat into the sky, or changing their shape or pressure rather than relying on gases. Balancing renewable energy output with demand is another crucial area for AI. Systems can combine various storage methods, with varying electricity supply to users who can be flexible – for example, cooling down buildings or making ice for thermal storage just ahead of peak demand. Electric vehicle charging can be synchronised with renewable output. Mainstream lithium-ion batteries could be supplemented or supplanted by varieties based on sodium or lithium metal, with solid electrolytes, or flow batteries with tanks of ionic chemicals. AI might also unlock currently less-favoured renewable options, such as finding good geothermal resources, which are independent of weather or time of day. But it seems increasingly likely, as I’ve observed previously , that a world of cheap, clean and abundant energy will attract greater demand. Data centres themselves have rapidly and recently emerged as a key source of growing energy needs. AI algorithms and chips will get more efficient and faster, but that will make their use more economically attractive. So, AI might take us down quite different energy paths from this current climate consensus, while still delivering much lower emissions. One of deep learning’s most famous achievements, AlphaFold, a subsidiary of Google parent Alphabet, won its inventors the Nobel Prize in Chemistry this year. It predicts the structure of proteins, a task that had proved beyond traditional models. DeepMind’s GNoME system, introduced last November, explores new chemical compounds, claiming to have found 2.2 million. Similar breakthroughs might deliver an energy system centred on atoms or molecules rather than electrons. The atom-based system is one centred on nuclear power. The surge in US data centre needs has already revived interest in mothballed and new reactors. Traditional reactors in the West have become too slow and costly to build. But China, South Korea and the UAE have managed to construct nuclear plants on reasonable budgets and timelines. New design and construction methods could revive nuclear power even in western countries. Its higher cost than renewables may be compensated by its geographic density, reliability and – cooling water needs apart – independence from weather and seasons. More excitingly, small modular reactors could, their proponents hope, be delivered more quickly and at lower costs, while also being inherently safer. They could even drive ships. And AI may hold its greatest promise in nuclear fusion – a source of limitless, concentrated and clean energy. Modelling and controlling the complex behaviour of plasmas at 100 million degrees Celsius and developing new materials that can cope with such extreme conditions has already benefited from advanced computing. Commercial fusion, often thought to be decades away, might come along sooner. Even if it doesn’t, a space-based human future from the second half of this century onwards will surely rely on it. Alternatively, the molecule-based system would combine traditional hydrocarbons, biofuels and biomaterials. Oil and gas production would stay competitive by becoming cleaner and more efficient, with intelligent systems seeking out hydrocarbons underground. Electrochemistry brings together the renewable or nuclear worlds with the provision and manipulation of molecules. Hydrogen from fossil fuels, renewable and nuclear electricity, or found naturally-occurring underground, would be converted to ammonia, methanol and synthetic fuels for planes, ships and heavy industry. Something like AlphaFold may discover new catalysts and genetic modification to turn sunlight into essential chemicals and plastics. Optimised carbon capture would ensure oil, gas and coal can be used without greenhouse gas emissions. Reducing carbon dioxide and methane already in the atmosphere is a critical task for any future energy-climate paradigm. The best methods for doing so might be chemical, electrochemical, biological or mineralogical. At Adipec, Adnoc announced promising progress at converting carbon dioxide into solid minerals underground in Fujairah. These broad pictures are, of course, simplistic. Any successful future energy system will use a much higher share of electricity than today, will be more efficient even as it uses more energy, and will mix electrons, atoms and molecules. Different regions will pursue distinct approaches depending on their legacy infrastructure, endowment of resources, economic needs and public opinion. An electron-based system still needs huge quantities of materials: metals such as copper, aluminium, lithium and rare earths, polymers, ceramics and novel stuff such as graphene. And AI’s assistance in finding brilliant technical solutions won’t be enough on its own, without contending with the complexity, messiness and reflexivity of humanity’s social, political and legal systems. Successful inventions aren’t necessarily just, nor acceptable to everyone. A conference of top business leaders, ministers, scientists, engineers and bright young students is a starting point in ensuring human intelligence makes best use of the artificial version. Robin M. Mills is chief executive of Qamar Energy, and author of The Myth of the Oil Crisis