AI’s Excessive Water Consumption Threatens to Drown Out its Environmental Contributions

By Joyeeta Gupta, Hilmer Bosch, Luc van Vliet 24 April, 2024

In this article from The Conversation, Gupta, Bosch & Van Vliet delve into the paradoxical nature of AI in addressing water challenges while it being a big water user itself. Will the contributions of AI be overshadowed by its huge water footprint?

Tech optimists argue AI holds potential to solve the world’s water problems but it uses a lot of water - ChatGPT uses 500ml for 5-50 prompts, way more than a Google search
There's also the related water pollution & the thirsty data centres that will need more cooling with climate change & are now even being placed in dry sub-Saharan Africa
There is sufficient evidence for concern that the rapid uptake of AI risks exacerbating the water crises rather than help addressing them

The article was first published in The Conversation in Mar 2024. Click here to view.

Water is needed for development, production and consumption, yet we are overusing and polluting an unsubstitutable resource and system.

Eight safe and just boundaries for five domains (climate, biosphere, water, nutrients and aerosols) have been identified beyond which there is significant harm to humans and nature and the risk of crossing tipping points increases. Humans have already crossed the safe and just Earth System Boundaries for water.

To date, seven of the eight boundaries have been crossed, and although the aerosol boundary has not been crossed at the global level, it has been crossed at city level in many parts of the world.

Both water targets (surface & ground) of the safe & just boundaries have been crossed

For water, the safe and just boundaries specify that surface water flows should not fluctuate more than 20 per cent relative to the natural flow on a monthly basis; while groundwater withdrawal should not be more than the recharge rate. Both of these boundaries have been crossed.

These thresholds have been crossed even though the minimum needs of the world’s poorest to access water and sanitation services have not been met. Addressing these needs will put an even greater pressure on already-strained water systems.

AI’s potential

Technological optimists argue that artificial intelligence (AI) holds the potential to solve the world’s water problems. Supporters of AI argue that it can help achieve both the environmental and social Sustainable Development Goals (SDGs), for example by designing systems to address shortages of teachers and doctors, increase crop yields and manage our energy needs.

In the past decade, research into this area has grown exponentially, with potential applications including increasing water efficiency and monitoring in agriculture, water security and enhancing wastewater treatment.

Tech optimists argue AI holds potential to solve the world’s water problems like smart irrigation…

…but unclear if AI can actually make a difference or if it exacerbates existing challenges

AI-powered biosensors can more accurately detect toxic chemicals in drinking water than current quality monitoring practices.

The potential for AI to change the water used in agriculture is evident through the building of smart machines, robots and sensors that optimize farming systems.

For example, smart irrigation automates irrigation through the collection and analysis of data to optimize water usage by improving efficiency and detecting leakage.

As international development scholars who study the relationship between water, the environment and global inequality, we are curious about whether AI can actually make a difference or whether it exacerbates existing challenges. Although there is peer-reviewed literature on the use of AI for managing water and the SDGs, there are no peer-reviewed papers on the direct and indirect implications of AI on water use.

AI and water use

Initial research shows that AI has a significant water footprint. It uses water both for cooling the servers that power its computations and for producing the energy it consumes. As AI becomes more integrated into our societies, its water footprint will inevitably grow.

AI uses a lot of water e.g. ChatGPT uses 500ml for every 5-50 prompts; much more than a Google search…

…it also pollutes water though related hardware production

The growth of ChatGPT and similar AI models has been hailed as “the new Google.” But while a single Google search requires half a millilitre of water in energy, ChatGPT consumes 500 millilitres of water for every five to 50 prompts.

AI uses and pollutes water through related hardware production. Producing the AI hardware involves resource-intensive mining for rare materials such as silicon, germanium, gallium, boron and phosphorous. Extracting these minerals has a significant impact on the environment and contributes to water pollution.

Semiconductors and microchips require large volumes of water in the manufacturing stage. Other hardware, such as for various sensors, also have an associated water footprint.

Data centres provide the physical infrastructure for training and running AI, and their energy consumption could double by 2026. Technology firms using water to run and cool these data centres potentially require water withdrawals of 4.2 to 6.6 billion cubic metres by 2027.

Water use of big tech companies’ data centres grossly underestimated…

…plus, water demand set to increase for cooling with rising temps from climate change

By comparison, Google’s data centres used over 21 billion litres of potable water in 2022, an increase of 20 per cent on its 2021 usage.

Training an AI at the computing level of a human brain for one year can cost 126,000 litres of water. Each year the computing power needed to train AI increases tenfold, requiring more resources.

Water use of big tech companies’ data centres is grossly underestimated — for example, the water consumption at Microsoft’s Dutch data centre was four times their initial plans. Demand for water for cooling will only increase because of rising average temperatures due to climate change.

Conflicting needs

The technology sector’s water demand is so high that communities are protesting against it as it threatens their livelihoods. Google’s data centre in drought-prone The Dalles, Ore. is sparking concern as it uses a quarter of the city’s water.

Taiwan, responsible for 90 per cent of the world’s advanced semiconductor chip production, has resorted to cloud seeding, water desalination, interbasin water transfers and halting irrigation for 180,000 hectares to address its water needs.

Locating data centres

As water becomes increasingly expensive and scarce in relation to demand, companies are now strategically placing their data centres in the developing world — even in dry sub-Saharan Africa, data centre investments are increasing.

Data centres being placed in developing countries – even dry sub-Saharan Africa

Google’s planned data centre in Uruguay, which recently suffered its worst drought in 74 years, would require 7.6 million litres per day, sparking widespread protest.

What emerges is a familiar picture of geographic inequality, as developing countries find themselves caught in a dilemma between the economic benefits offered by international investment and the strain this places on local water resources availability.

“We believe there is sufficient evidence for concern that the rapid uptake of AI risks exacerbating the water crises rather than help addressing them”

We believe there is sufficient evidence for concern that the rapid uptake of AI risks exacerbating the water crises rather than help addressing them. As yet, there are no systematic studies on the AI industry and its water consumption. Technology companies have been tightlipped about the water footprint of their new products.

 

The broader question is: Will the social and environmental contributions of AI be overshadowed by its huge water footprint?

Further readings

More on Latest

Author: Joyeeta Gupta
Joyeeta Gupta is the 2023 winner of the Spinoza Prize, the highest prize in the Netherlands for academic work (Euro 1.5 Million). She co-chaired UNEP’s Global Environment Outlook-6 (2016-2021), published by Cambridge University Press, presented to governments at the United Nations Environment Assembly in 2019, and won the Association of American Publishers PROSE award for Environmental Science. She is presently cochair of the Earth Commission (2019-2022), set up by Future Earth, together with Johan Rockström and Dahe Qin and has a plenary presentation to the World Economic Forum in Davos in 2023; and is member of the Global Commission on the Economics of Water (2022-2024) which is preparing a report for the UN Water Conference of 2023. In 2022, she was awarded the 2022 Piers Sellers Prize for world leading contribution to solution-focused climate research, recent world-leading research outputs and evidence of resulting impacts; application of an interdisciplinary approach; and focus on climate solutions, Priestley International Centre for Climate, Leeds University. She is full professor of environment and development in the global south at the University of Amsterdam and IHE Delft Institute for Water Education. She is also the Faculty Professor on Sustainability (2019-2024). She leads the programme group on Governance and Inclusive Development. She was lead author in the Intergovernmental Panel on Climate Change which won the 2007 Nobel Peace Prize with Al Gore and of the Millennium Ecosystem Assessment which won the Zaved Second Prize. She has published several books, is on the editorial board of seven journals, and has (co-)edited 13 Special Issues. Her Google Scholar Impact Factor is 64, with 120 journal papers and more than 17,000 citations. She has successfully supervised 29 PhDs and is currently supervising 20 PhD students in the areas of climate change, forest, food/fish, water and disaster governance as well as in development challenges such as food governance and child marriage. She has been on the scientific steering committees of international and national scientific programmes. At national level, she was the Vice-President of the Commission on Development Cooperation (2011-2019) and member of the Advisory Council on International Affairs (2011-2019), a statutory body that advises three Cabinet Ministers in the Netherlands. In 2019 she won an ERC Advanced Grant of 2.5 million Euros for work on climate change and fossil fuels.
Read more from Joyeeta Gupta →
Author: Hilmer Bosch
Dr Hilmer Bosch is a postdoctoral Researcher with the Global Commission on the Economics of Water at the University of Amsterdam. He holds degrees from the University of Strathclyde (MSc Environmental Entreperneurship), IHE-Delft Institute for Water Education (MSc Water Management and Governance) and a PhD from the Amsterdam Institute for Social Science Research at the University of Amsterdam. His unwavering passion lies in conducting research and shaping policies related to inclusive development, water justice and water property rights with a specific focus on the Global South.
Read more from Hilmer Bosch →
Author: Luc van Vliet
Luc van Vliet is a researcher with a passion for environmental governance, political and queer ecology, and climate and water justice. He has completed academic degrees in human geography, politics, philosophy, and environmental studies from Australian and Dutch universities. Currently, he work as a researcher for the Global Commission on the Economics of Water (GCEW).
Read more from Luc van Vliet →
«
»
Related Posts