AI's Growing Thirst for Energy and Water Sparks Environmental Concerns

Artificial intelligence and the data centers that power it are placing significant strain on global energy and water resources, prompting experts to call for reduced usage and greater transparency from AI companies. The increasing reliance on AI for everyday tasks, from simple calculations to generating content, contributes to a growing environmental footprint that runs counter to global decarbonization efforts.

“AI is going in the opposite direction to decarbonization efforts,” said Sasha Luccioni, co-founder and chief scientific officer of the Sustainable AI Group. “We should be thinking about where we are going towards. If you’re recycling and a vegan but then you’re using ChatGPT to do your multiplication for you, well that’s kind of against the trend.”

Experts suggest that the most effective way to mitigate AI's environmental impact is to use it less frequently. Kaveh Madani, director of the United Nations University Institute for Water, Environment and Health, stated, “The cleanest form of AI use is no use. So when you could avoid using AI, don't use it.” This advice extends to avoiding AI for tasks like calculations, directions, store hours, recipes, or shopping lists, which were historically managed without AI.

Luccioni noted that while AI can offer productivity gains, its current application for many day-to-day tasks is excessive. “You really don’t need all of these generative AI technologies to do day-to-day tasks,” she said. “I do agree there are some productivity gains to be had but I think that it’s a pretty small percentage of what people are currently using.”

Making queries more concise can also reduce resource consumption, as more information requires more computing power, energy, and water. Experts recommend being direct and avoiding unnecessary politeness or background details in AI prompts.

Escalating Resource Demands

Global data centers consumed an estimated 448 trillion watt-hours of electricity last year, a figure surpassing the consumption of all but ten countries. This demand is projected to more than double in the next four years, potentially ranking data center electricity usage behind only five countries. By 2030, the water required to cool these data centers alone could reach approximately 9.3 trillion liters annually, enough to supply the world's drinking water needs for 1.7 years.

Generating a text response from AI is akin to running an efficient light bulb for two and a half minutes, but with ChatGPT alone, this action is performed an estimated 2.5 billion times daily. Generating complex AI video content can consume the equivalent of 42 hours of light bulb usage and one gallon of water.

Lack of Transparency and User Control

Private AI companies have largely failed to provide transparent data on the energy and water costs associated with their services. This lack of disclosure forces researchers to rely on estimates based on open-source AI models. Mosharaf Chowdhury, a computer science professor at the University of Michigan who studies AI energy consumption, stated, “We have no way of knowing and getting a sense of the amount of energy.”

Ana Pinheiro Privette, a former sustainability official for Amazon Web Services, emphasized the issue of user control. “If there’s no transparency, we have no choice. We’re really not choosing. We are being given whatever is being given to us,” she said. “That’s the power. The power is to say ‘I actually want to understand what I’m consuming’.”

Many search engines, including Google, integrate AI-powered responses without explicit user requests, effectively requiring users to opt out of AI rather than opt in. Luccioni suggested that users can opt out of AI features in Google searches by appending "-ai" to their query or by selecting the "Web" option in search results. Alternative search engines like Ecosia, which uses less energy in its AI and plants trees, or DuckDuckGo and Startpage, which offer no-AI options, are also available.

Growing Community Opposition

The growing awareness of AI's environmental impact is also leading to increased community opposition to data center development. Balaji Tammabattula, chief operating officer of BaRupOn, noted that "the moment you say that you’re building a data center, there’s a backlash. The data center is the new boogeyman.” He added that companies must listen to community concerns and work towards more sustainable practices. "AI is not going anywhere,” Tammabattula said. “It has to be done. But it has to be with the help of the community, where we're understanding the concerns of the community.”

Privette believes that consumer awareness can influence corporate behavior. “The big power I think the consumer has is the market message because I’ve seen that when I worked at Amazon,” she said. “They listen. They listen if everybody suddenly starts caring about not having a footprint.”