The explosive growth of AI technology is fueling massive demand for data centers, and to meet the need, companies and investors are looking everywhere—from space to the bottom of the ocean and even California fairgrounds—for a possible solution.
Cassie Buhler (center), a postdoctoral fellow who co-authored the commentary in Nature that calls for the expansion of open source AI models amid growing concerns about AI energy and resource use. Credit: Lauren Lipuma/CIRES.
But in a new commentary published today in Nature, Associate Professor Carl Boettiger and collaborators at the Eric and Wendy Schmidt Center for Data Science & Environment (DSE) argue that local, open models could reduce AI’s environmental footprint while improving privacy, access, and scientific usefulness. These free-to-use models—which can be downloaded and run entirely on personal computers—allow for greater control over user data and may use up to 100 times less energy than commercial counterparts like ChatGPT or Claude.
“Many climate scientists have avoided AI altogether because of the environmental and privacy costs of relying on data centers, and we share those concerns,” said Boettiger, who is a faculty advisor to the Schmidt DSE and member of the Department of Environmental Science, Policy & Management. “Yet when AI is used responsibly, we also recognize that it can be a powerful tool for solving problems.”
In an accompanying Q&A, Boettiger and his co-author Fernando Pérez, faculty co-director of Schmidt DSE and associate professor of statistics, describe how smaller, open-source large language models are rapidly catching up to the capabilities offered by enterprise models reliant on data centers. They believe that, by embracing these open-source models, scientists and the public have the opportunity to create AI technologies that are better suited to their needs while providing alternatives to large, privately owned platforms.
Read More
-
Why scientists should lead the shift away from AI mega data centres (Nature)
-
AI Can Move Beyond Data Centers (Schmidt DSE)
-
Why we don’t need more data centers to build better AI (Berkeley News)