Sustainable futures
How the UK can take the lead in sustainable AI
Artificial Intelligence can be leveraged to accelerate progress towards net zero. Yet the environmental cost of AI has yet to addressed by policymakers, says Professor Tom Rodden.
AI holds great promise for net zero, but Its environmental costs are rising fast
There are high hopes that AI can help tackle some of the world’s biggest environmental challenges. But there’s a negative side to the explosion of artificial intelligence and its infrastructure. AI servers are thirsty consumers of water and use massive amounts of electricity, as well as demanding huge quantities of rare materials for computing hardware.
The United Nations Environment Programme (UNEP) cites a study estimating that global AI-related infrastructure may soon consume six times more water than Denmark, a country of six million. In Ireland, an emerging AI tech hub, AI data centres could account for nearly 35% of the country’s energy use by 2026. The UK is already Europe’s largest data centre market and the National Grid predicts that the electricity demand from UK data centres will increase sixfold by 2034, to 30% of total electricity used.
The demand for increasing amounts of computational power is unprecedented, which has huge implications for resources, infrastructure and the environment. Yet while it is reassuring that the safety and security of AI is debated, there is less focus on monitoring its environmental impact, with calls instead to cut red tape and regulation to unleash the potential of this booming industry.
Without regulation and transparency, we are flying blind on the environmental impact of AI
As a computer scientist, I’m excited by harnessing the power of AI to accelerate positive change. It’s already accelerating drug discovery, patient diagnosis and treatment and optimising energy consumption. At the University of Nottingham, our pioneering use of AI for short-term weather forecasting is making solar power more efficient, and reducing the carbon footprint of airports like Heathrow by optimising air traffic operations.
But there is a compelling case for environmental regulation for AI
I am chairing the working group behind a Royal Academy of Engineering report that calls on the UK government to take the lead in ensuring the net effect of AI on the planet is positive, and for tech companies to accurately report how much energy and water their data centres are using. Every data centre has a metering system to record energy and water use – but there’s no obligation to make this data available to policymakers. So, ministers and local authorities are playing a guessing game, whether trying to assess the environmental impact of siting a data centre in your town, or the impact on grid capacity and access to renewable energy for businesses and communities in ‘AI Growth Zones’.
Data centres can be designed to use less energy, drinking water and critical materials. Researchers at the University of Nottingham and our partners are supporting this, such as by developing ultra-low power, atomically-thin semiconductors, which will reduce the electricity demand from AI data centres and high-performance computing by more than 90%.
But rolling out such technological advances at scale requires evidence on the use of resources by data centres and of the environmental cost.
Measuring the footprint of AI is essential to building a sustainable digital future
Effective environmental regulation also plays to the UK’s strengths. Today, Britain is the third largest AI market in the world and we are acknowledged leaders in AI assurance technologies. The UK AI assurance market has the potential to reach £6.53bn by 2035 – a growth of over £5bn from today. Environmental monitoring will not hinder opportunities for growth as the government ramps up AI adoption across the UK. By working with regulators, we can create models for environmental compliance that don’t burden industry or hinder innovation. Rather, it will allow policymakers to mitigate risks and map out the implications for resources and infrastructure, and shape development to suit the UK’s needs.
Demand for AI will grow exponentially and it is vital we can measure the impact of the infrastructure it needs. The Engineering Responsible AI report also calls for the embedding of sustainability into government’s approach to investing in AI, and for all stakeholders across the value chain to have a clear understanding of the environmental impacts of AI systems and services. Once we have access to evidence on the environmental impact of data centres, we can plan a sustainable future for AI in the UK.
Finally, we may all need to think about our use of AI and the resources we are consuming when we use AI tools. One estimate suggests that asking ChatGPT to generate a 100-word email consumes 0.5 litres of water. This would include the AI summary you see by default at the top of some Google searches.
As AI continues to evolve and expand, so too must our understanding of its environmental footprint. By embedding sustainability into AI development and demanding transparency from industry, we can ensure that this powerful technology supports, not undermines, our path to a greener future. The choices we make now will shape not only the future of AI, but the future of the planet.
This article was written by Professor Tom Rodden, and it forms part of the Institute for Policy and Engagement’s ongoing campaign highlighting research that showcases how the University of Nottingham is advancing the Sustainable Development Goals (SDGs) at different levels.
Tom Rodden
Tom Rodden is Pro-Vice-Chancellor of Research and Knowledge Exchange and Professor of Computing at the University of Nottingham. He is Chair of the working group behind Engineering Responsible AI, a report developed by the Royal Academy of Engineering in partnership with the Institution of Engineering and Technology and BCS, the Chartered Institute of IT, under the National Engineering Policy Centre.