The Environmental Consequences of AI

Artificial intelligence may appear to be computerized magic, boundless in its ability to improve productivity, advance scientific research, create art, and more. But AI comes with a substantial environmental cost. Recognizing its toll may inspire more-responsible use.

Generative-AI data centers require "a staggering amount of electricity," according to MIT News. To meet demand, U.S. utility companies reportedly plan to add 133 new natural gas-fired power plants to the nation's grid, as well as additional oil- and coal-fired plants. These new power plants will come online during a critical time for slowing climate change and may threaten "decades of progress cutting greenhouse gas emissions," according to a Washington Post article.

Additionally, data centers that power AI systems require massive amounts of water for cooling: A University of California, Riverside, study found that writing a 100-word email using ChatGPT consumes the equivalent of just over one standard bottle of water (16.9 fl. oz.). The study's authors note that freshwater scarcity has become a modern-day challenge, and the global AI demand for cooling water in 2027 alone is projected to reach the equivalent of about half of all the water used in the United Kingdom annually.