Artificial intelligence is often blamed for increasing energy use and exacerbating climate change, but new research suggests that its overall impact on global emissions is surprisingly small. The findings point to potential environmental and even economic benefits as AI continues to expand.
Researchers from the University of Waterloo and Georgia Tech analyzed data from across the U.S. economy, along with estimates of how widely AI is being used in various industries. Their goal was to understand what would happen to energy use and emissions if AI adoption continued to increase at its current pace.
According to data from the U.S. Energy Information Administration, 83% of the U.S. economy remains dependent on fossil fuels such as oil, coal, and natural gas, all of which produce greenhouse gases when burned. Researchers found that AI-related electricity usage in the United States is comparable to the total energy consumption of Iceland. Still, this increase is too small to have a significant impact on national or global emissions.
Local effects near data centers
“It is important to note that the increase in energy use will not be uniform; the increase will be felt more where electricity is generated to power data centres,” said Dr. Juan Moreno-Cruz, professor in the School of Environment at the University of Waterloo and Canada Research Chair in Energy Transition. “If you look at that energy from a local perspective, this is a big problem because in some places, power generation and emissions could double. But on a larger scale, the use of energy by AI will be invisible.”
Although this study does not focus on how local economies are affected in areas with high concentrations of data centers, the broader findings remain encouraging.
AI as a potential solution to climate change
“We are offering a different perspective to those who believe that the use of AI will be a big problem for the climate and that we should avoid using it,” Moreno-Cruz said. “The impact on the climate is modest, and AI can be used to develop green technologies or improve existing technologies.”
To reach these conclusions, Moreno-Cruz and fellow environmental economist Dr. Anthony Harding evaluated different sectors of the economy, the types of jobs in those sectors, and how many of those roles could be handled by AI.
The researchers plan to expand their analysis to other countries to better understand how the introduction of AI may impact energy use and emissions around the world.
The study, “Watts and Bots: The Energy Impact of AI Deployment,” environmental research letters.
