The environmental impacts of artificial intelligence go far beyond energy, with potentially severe effects on water, land use and carbon pollution, according to a new analysis by the United Nations.
The report looks at data centers and supply chains supporting AI and quantifies the amount of water and land needed to generate the electricity for powering those facilities around the globe. It also examines the amount of carbon dioxide that could be generated by AI’s growing energy demands.
A key theme of the analysis is that AI stands to have large environmental consequences, even if renewable energy is used to power data centers. Low-carbon electricity, for instance, can consume large amounts of water or land. Using bioenergy rather than coal, for example, cuts the amount of carbon pollution significantly but is far more water intensive.
“Evaluating sustainability through a single metric can hide trade-offs and shift burdens onto places already facing water stress or land pressure,” the report says.
