Artificial intelligence: climate solution or climate catastrophe? | Green Edition 2025

With the increasing prevalence of artificial intelligence (AI) in education, science, and research, academia is rapidly determining whether the technology will serve as an asset or a threat. Similarly, in regards to climate change, some believe AI will miraculously solve the crisis, and others believe the emissions impact is just too costly.

AI requires massive amounts of energy to operate. Data centers, which physically store information, currently account for 1% to 1.5% of global electricity consumption. This number will likely rise with increased operation rates.

“Projections from the U.S. Department of Energy have changed because of the rise in generative AI and other data center infrastructure,” UW PhD student Amelia Doğan said.

Regions with cleaner energy grids have a lower emissions impact. However, powering data centers with renewable energy presents several barriers. AI, especially generative AI, requires a consistent source of a lot of energy. Additionally, renewable sources like solar and wind rely on certain conditions, making it difficult to ensure a stable power supply necessary for data centers.

“Renewables are more prone to fluctuations,” Doğan said.

Furthermore, this transition is hindered by the federal government’s recent shift away from renewable energy. As a solution, corporations that have their data stored in different regions have been able to predict when energy will be available and assign workloads accordingly.

“You could schedule computation[s] to happen when renewables are available,” Vikram Iyer, UW assistant professor of mechanical engineering, said.

In addition to energy consumption, data centers also require huge amounts of water for cooling.

“They have to often strike special deals with water providers to make sure that they’re getting enough water,” Doğan said.

Additionally, hardware production for these data centers requires resource extraction, which often pollutes water sources and contributes to electronic waste after disposal.

Indirectly, AI has also increased energy consumption by influencing human behavior, such as creating algorithms that ultimately increase online purchasing.

However, artificial intelligence models have the potential to optimize research processes and data collection. AI is very adept at recognizing patterns and anomalies. For that reason, it has been leveraged to aid a myriad of monitoring projects, making it easier for researchers to track emissions of specific institutions, wildlife conservation efforts, and weather patterns. Non-profit organizations like ClimateTRACE implement AI and remote sensing to monitor greenhouse gas emissions in different regions. AI has also been utilized to increase the efficiency of renewable energy sources through predictive maintenance.

Simultaneously, AI has been used to optimize mining and drilling processes for fossil fuel corporations, strengthening its competitiveness against renewable alternatives.

The precise carbon footprint of artificial intelligence is difficult to calculate, partially due to a lack of transparency at the corporate level. Resolving this would require further policy and regulation of emissions for tech companies.

At UW, professors, faculty, and students are researching and working on all angles of AI integration, namely from the departments of engineering, policy, and atmospheric sciences.

Reach writer Abigail Meyers at specials@dailyuw.com. X: @abby_meyers05

Like what you’re reading? Support high-quality student journalism by donating here.