What impact will artificial intelligence have on the energy sector?

Artificial Intelligence: A solution to new energy problems that it partly created itself.

“It is difficult to overestimate the potential of artificial intelligence in the fight against climate change,” said the Minister of Industry and Advanced Technologies and Adnoc Chief Executive Dr Sultan Al Jaber recently wrote on Project Syndicate.

However, the sector’s huge electricity consumption is straining the grid and risks increasing coal and gas consumption, and therefore greenhouse gas emissions.

The impact of artificial intelligence on the energy sector will be revealed in three periods.

At least in the short term, this will increase energy demand. And if low-carbon energy and efficiency can’t keep up, that inevitably means more greenhouse gas emissions. AI’s computations currently account for less than 1 percent of global emissions—that could grow significantly, but by how much depends on how quickly it develops and the choices we make.

Google aims to achieve net zero carbon emissions by 2030.

But emissions have risen by 48 per cent since 2019 – and much of that is due to the expansion of data centres and artificial intelligence. The 14.3 million tonnes of carbon dioxide emitted last year was more than the entire emissions of the Baltic nation of Estonia. That’s despite running 64 per cent of its operations on low-carbon electricity – using renewable and nuclear power.

In the medium term, AI promises untold benefits in terms of the volume, efficiency and cleanliness of energy production and use.

Dr. Al Jaber’s commentary cites the benefits of balancing variable renewable energy generation, identifying molecular structures that trap carbon dioxide, reducing water use while increasing crop yields, and “breakthroughs in fusion, hydrogen, and modular nuclear power (as well as) long-term battery storage.”

In November, he convened the Change Makers Majlis conference in Abu Dhabi to discuss artificial intelligence and energy transformation.

A team at the Innovation for Cool Earth Forum, chaired by my Columbia University colleague David Sandalow, has outlined other possibilities in detail.

The more spectacular capabilities fall into three general categories: optimizing and automating complex systems; discovering patterns and relationships in large data sets; and designing and simulating technologies that would otherwise be impossible to implement.

In the first group are opportunities such as matching renewable generation with weather forecasting, demand, home heating and cooling, battery storage and electric vehicle charging. Reducing the financial and psychological burden of cutting emissions and saving energy costs could help ease some of the political resistance to net-zero policies.

The second group includes studies that combine the vast published literature in a way that is beyond the capabilities of any single scientist. AI systems can test a large number of designs for advanced batteries, carbon capture materials, or catalysts for biofuels or hydrogen, and select the most promising ones for further development. Already in 2022, Deep Mind, a subsidiary of Google’s parent company Alphabet, unveiled a system for predicting the structure of 200 million proteins.

Such systems can monitor global greenhouse gas emissions and carbon dioxide storage in near real time. They can search large data sets of remote sensing, geochemistry, surface seismic, magnetic, radioactive and gravity readings to find new deposits of hydrocarbons and key energy minerals such as rare earths, uranium, lithium and copper.

KoBold Metals, a startup backed by Bill Gates and Jeff Bezos, mining company BHP and Norwegian state oil giant Equinor, says it has found a large copper deposit in Zambia using artificial intelligence that the country’s president says could become one of the world’s three largest.

The third area could include things like simulating nuclear fusion. This inexhaustible source of clean energy requires us to use a plasma at a hundred million degrees, confine it in intense magnetic fields, and keep the reaction stable without melting the chamber containing it. Models could predict room-temperature superconductors that would eliminate losses in electrical transmission.

AI computing will also become more energy efficient. The combination of such capabilities means that in the medium term, energy use will be better targeted, supply will be increased and reduced cost-effectively, and emissions will potentially fall significantly – with the right policy choices.

Beyond these, in the long term – the 2050s and beyond – there are things that are intangible.

We often hear that all it takes is “political will” to beat climate change and that we have all the technology we need. Instead of delivering the dark arts of electoral manipulation spiced up with big data from social media, could AI simulate entire societies and economies to help design and implement climate-friendly policies that gain widespread acceptance?

Some of the more ambitious goals for AI are that by the 2030s, intelligence will surpass humans in almost every way.

Would such a creature prioritize highly effective climate action? Or would it reflect the attitudes of its creators, perhaps a trillionaire, corporation, or country that cares little for the well-being of billions of vulnerable people?

But we can be sure of one thing: this AI world will require much more energy than today. It could be superintelligence, humanoid robots, global climate purification systems, simulations of entire societies, extraterrestrial industry, extraplanetary colonization or other unexpected events. The trend of technology is always to require more and better energy, even as it becomes more efficient and cleaner.

Gulf countries could be among the biggest beneficiaries: they have money, growing technical expertise and a focus on artificial intelligence that could solve many of their key problems.

Relatively cheap, clean, and abundant energy will help build computing power, partially offset by the greater need for cooling compared to temperate climates.

With relatively small populations relative to their economies and energy resources, they will benefit disproportionately from multiplying the efficiency of their people. Adnoc has taken steps in this regard through its AIQ joint venture with Abu Dhabi-based AI firm G42 and through its cross-sharing with big data analytics firm Presight.

Success depends on gaining expertise, combining AI with the best of human intelligence, and reaching and expanding the frontiers of knowledge.

It also requires much more and better data for training systems. This is one area where the Gulf needs to improve: rather than hoarding information to preserve private empires in the name of national security, data from the energy sector and many other areas must be accessible, reliable, transparent and organized.

Competitors – whether competing energy producers or competing sources and methods of obtaining energy – do not sit idly at their keyboards.

At the moment, artificial intelligence means an improvement in the industry and an opportunity to sell energy.

But in the long term – whether it’s 30 years from now or 30 months from now in the world of AI – this will change the energy industry.

Robin M. Mills is CEO of Qamar Energy and author of The Myth of the Oil Crisis

Updated: July 8, 2024, 3:00 AM