AI is an energy and water hog, here’s what you can do to counter that
Artificial intelligence and its supporting data centers are consuming increasing amounts of energy and water, exacerbating climate change, with companies often lacking transparency about their resource use. Experts advocate for individuals to reduce their AI usage and for companies to be more transparent. Data center developers acknowledge growing community opposition and emphasize the need for sustainable practices.
Artificial intelligence and the data centers required to power it are significantly increasing their energy and water consumption, moving in the opposite direction of global decarbonization efforts, according to experts. Cognitive computer scientist Sasha Luccioni, co-founder of the Sustainable AI Group, notes that every online query contributes to this environmental footprint, with AI companies often lacking transparency regarding their resource usage. This trend is problematic, especially as generative AI is increasingly integrated into everyday tools, making it harder for users to opt out.
Experts like Kaveh Madani, a water scientist with the United Nations University Institute for Water, Environment and Health, suggest that individuals can counter this by simply using AI less often, particularly for simple tasks that don't require generative AI. They also recommend making queries concise to reduce computing and resource demands. A new report from the United Nations University projects global data center electricity use to more than double in four years, potentially requiring 2.5 trillion gallons of water annually by 2030, enough drinking water for the entire world for 1.7 years.
The lack of transparency from private AI companies regarding their energy and water costs makes it challenging for researchers to accurately track these impacts, forcing them to rely on estimates from open-source AI models. Mosharaf Chowdhury, a University of Michigan computer science professor, and Ana Pinheiro Privette, a former Amazon Web Services sustainability official, highlight the lack of consumer control, with AI often integrated by default into search engines like Google.
Privette emphasizes the power consumers have through market messaging, noting that companies listen when there's collective concern. Balaji Tammabattula, COO of BaRupOn, an energy-ready data center campus developer, acknowledges the growing community backlash against data center construction, describing them as "the new boogeyman." He stresses that while AI development is inevitable, it must proceed with community understanding and through more sustainable, energy- and water-efficient practices, citing examples like data centers in two Virginia counties that used 2.1 billion gallons of water in 2023.