AI Power Usage

To power and train artificial intelligence tools, the use of data centers has led to increased usage of natural resources, including but not limited to fresh water, minerals, and fossil fuels. Fresh water and air turbines are two ways that data centers use to cool down the computers which are handling great amounts of processing power.

Fresh Water

For AI, fresh water is used to provide cooling, water-intensive electricity generation, and to create materials throughout the AI supply chain. In data centers, fresh water is used in cooling systems to cool the hundreds or thousands of computers generating computational heat. The fresh water used does not return to the local watershed but is recycled until it evaporates. Water-intensive electricity generation is a term that references the many methods using water to create electricity, such as dams and reservoirs. As the size and complexity of AI models becomes great, the amount of water usage increases. 

Fossil Fuels & Rare Earth Materials

The physical components of AI require the use of rare earth materials such as lithium, copper, and cobalt, which are used to create batteries, wiring, and more. Mining these minerals can cause detrimental results to the environment from deforestation to soil erosion. Powering AI data centers require large amounts of electricity and water; the computer processors require fans or water to stay cool, resulting in water and electricity consumption that can strain local communities. Electricity is often made by burning coal and natural gas which results in pollutant emissions in addition to the consumption of these earth materials. 

What this means for you

-Consider each usage of AI models and whether another non-generative AI model can be used. Be conscientious of your use of AI tools.

Additional Information: 

AI's Challenging Waters by Anna Pinheiro Privette 

Why the true water footprint of AI is so elusive by Emily Kwong, Regina G. Barber, Hannah Chinn, Rebecca Ramirez (NPR)

Explained: Generative AI's Environmental Impact by Adam Zewe (MIT News)

Can Data Centers be Sources of Sustainable Heat? by Visual Capitalist

Can We Mitigate AI's Environmental Impacts? by Yale School of the Environment

Why AI uses so much energy - and what we can do about it by PennState Institute of Energy and the Environment

Teaching AI Ethics by Leon Furze