Taking the pulse of our planet with AI
By Thomas Lamanauskas, Deputy Secretary-General, ITU
Our planet is on fire – and flooded by water. With each passing month, we keep breaking the wrong kind of records.
April 2024 was the eleventh in a row with record-high global temperatures. The month was 1.58 degrees Celsius warmer than the estimated April average before global industrialization.
The consequences include forest fires from Canada to Chile, a drought in the Amazon rainforest, and severe floods from Afghanistan to Kenya.
The most vulnerable communities are often the hardest hit. Rising ocean levels, for example, have put Small Island Developing States in severe peril.
When it comes to the climate crisis, is artificial intelligence (AI) part of the problem?
Or will it help us find a solution?
AI is hungry and thirsty…
For now, our AI use leaves a growing carbon footprint.
These technologies are hungry for electricity – more than they can get from renewable sources.
Two years from now, the data centres that support skyrocketing AI use could consume twice as much energy as Japan does today, according to the International Energy Agency.
A recent study by the International Telecommunication Union (ITU) and the World Bank shows electricity use for data centres increasing by 57 per cent in the two years up to 2022.
Greenhouse gas (GHG) emissions from data centres have reached 32 million tonnes in 2022, up 45 per cent from two years earlier.
AI developers, users, and policy makers – all of us – need to bring this under control.
To hold the line against global warming, we need to rein in exponential emissions growth, put the whole industry on a better trajectory, and cut tech-related emissions to net-zero by 2050.
The necessary path is captured in the key ITU standard L.1470 – starting with a crucial 45 per cent cut in digital industry’s emissions in our present decade.
Of course, AI is thirsty, too, with researchers finding that just 10 prompts can consume as much as half a litre of water.
It also brings the risk of intensifying resource competition – as competing AI hubs scramble for energy supplies, rare minerals, and scarce microprocessor chips.
…but also a powerful tool for climate action
The same technologies can also bolster eco-solutions and help protect biodiversity. For example, AI systems can detect and analyse subtle eco-system changes and bolster conservation efforts.
AI solutions are renowned for their ability to boost energy efficiency and cut waste and emissions across other sectors. Recent research suggests AI can help mitigate between 5 and 10 per cent of global GHG emissions by 2030 — equivalent to the total annual emissions of the European Union.
Big data and AI can support smart energy control for telecom sites and data centres – especially by prioritizing renewable and low-carbon power sources.
On balance, the potential for AI-based climate solutions is crystal clear. But this will be no help until we tackle the risks of AI growth head-on.
Committing to Green Digital Action
Importantly, the tech industry recognizes its responsibility.
Digital companies are leaders in renewable energy uptake, accounting for 60 per cent of global purchases by 2021. This has helped keep tech-related emissions relatively constant despite the surge of generative AI.
For several years, ITU standards have provided guidance on cutting tech-related emissions, as well as how to handle disused electrical and electronic equipment, or e-waste.
With our green standards package, we aim to rein in our industry’s energy and carbon footprint.
The Green Digital Action initiative launched ahead of last year’s climate conference, COP28, mobilized over 40 partners, including governments, businesses, and civil society, as well as UN agencies and international development banks.
ITU and partners aim to put digital solutions at the forefront of climate action. One way is by gathering better data and incentivizing, through transparency, the “greening” of the digital economy.
We aim to ensure that all emissions stemming from digital products and services are properly measured and reported.
Partners in Green Digital Action have pledged to publicly report their own climate data – a crucial step to get tech on the right side of history.
Similarly, we and our World Standards Cooperation partners have pledged to deliver technical standards that make both business and environmental sense.
We are encouraging the entire global tech industry to get on board. I encourage everyone to support Green Digital Action and help make AI part of the solution.
Ultimately, I am confident AI will help us respond to the climate crisis. But we must work hard – and work collaboratively – to make that happen.
Together, we can harness AI tools to save our planet – not sink it further.
Based on remarks by Tomas Lamanauskas at the AI for Good Global Summit on 31 May 2024.
