How AI can help and hurt the environment

Information technology’s global use of energy has stayed at around 1% for several years, a surprising statistic given how important the sector and its use of data centers has become.

But that may change in the coming years as companies and governments start to rely more on artificial intelligence, which could consume even more energy and resources with its higher computing power.

So while the data centers needed to support AI technology present a tremendous economic development opportunity for state and local governments, its energy-intensive nature may put an even greater strain on them than they are already experiencing. That reality may demand a more measured approach from state and local leaders looking to take advantage of the technology and meet their climate goals.

According to some estimates, increased AI use could double data center electricity demand by 2026. The amount of water needed to cool data centers could also spike, affecting water supplies. Meanwhile, electronic waste could grow as the chips needed to run AI software are created and then discarded once they’re used up.

“We're trying to be diligent; we're trying to be thoughtful,” said Boston Chief Information Officer Santiago Garces during the recent Route Fifty Innovation Spotlight on leadership in AI. “The use of these tools is expensive. These tools can be expensive from an environmental standpoint as well. They require a lot of energy. Energy requires cooling, which means they also require water.”

But the true impacts will be unknown for a while. In an interview, Benjamin C. Lee, professor of electrical and systems engineering and computer and information science at the University of Pennsylvania, noted that tech companies have so far achieved “economies of scale” with their data centers, maintaining energy efficiency and keeping costs down.

“Generative AI models, like ChatGPT, are going to be much more computationally expensive than traditional data center jobs,” he said. “For instance, if a Google web search consumes some amount of energy, asking ChatGPT to give you an answer might consume 7 to 10 times more energy. If everyone starts using GPT like they use the search engine, we will see a big increase in energy costs.”

Despite those environmental challenges, some leaders see a potential upsidet. Katy Ruckle, Washington’s chief privacy officer, said during the Route Fifty event that the technology has some “exciting use cases.” She pointed to experiments using AI for wildfire detection or screening seagrass, which are both “very specific to some of the environmental considerations” in the state.

The issue has caught some lawmakers’ attention in Congress, too. In February, a bicameral group of Democratic elected officials introduced the Artificial Intelligence Environmental Impacts Act, which would direct the National Institute of Standards and Technology to develop standards to measure and report how artificial intelligence affects the environment.

The legislation, which was referred to the Senate Commerce Committee to be discussed and debated, also would have the institute create a voluntary framework for AI developers to report environmental impacts andrequire an interagency study to investigate its positive and negative affects.

“There is a Dickensian quality to the use of AI when it comes to our environment: It can make our planet better, and it can make our planet worse,” Massachusetts Sen. Ed Markey said in a statement. “The development of the next generation of AI tools cannot come at the expense of the health of our planet.”

Penn’s Lee said having AI rely more on sustainable energy sources is a good path forward, although he acknowledged we are “far away” from the technology being carbon neutral. Cary Coglianese, a professor of law and political science at the university, said relying on more sustainable sources of energy to power AI will “go hand in hand” with the overall transition away from fossil fuels underway in the wider economy.

“GPT promises massive increases in productivity,” Lee said. “It’s a tool that will transform the way we live and work, so I do think that AI should be given a little bit of room to run before we say we need to bring in carbon neutrality to AI computing.”

Having reporting requirements for AI’s environmental impact also could help leaders understand how energy and other resources are being used. In statements released by Markey’s office to coincide with the bill’s introduction, various environmental groups said such transparency is key to understanding any harm AI may do.

“One of the many unknowns about AI is its impact on the environment,” Kaili Lambe, policy and advocacy director at nonprofit Accountable Tech, said in a statement. “As rapid advances have led to increasingly widespread use of LLMs and other data-intensive AI systems, it’s imperative that we understand and take steps to mitigate unintended harm, including the potential for environmental degradation. Too often transparency takes a backseat when it comes to Big Tech innovation.”

For state and local governments, then, they will need to navigate how to innovate and experiment with new technologies, while ensuring they do not have negative consequences on the environment and their ambitious goals to combat climate change. Garces said it means “finding the balance between public use and cost.”