Artificial Intelligence, For Better or Worse

If you’re worried about AI, you’re in good company. But most of us haven’t let our concern over AI drive us as far as Karen Boyd has. 

“I got a Ph.D. in this because I am worried about it. That's literally why I chose to do that,” says Dr. Boyd, an AI researcher with the Policy & Innovation Center, a San Diego–based think tank focused on social, economic, and environmental problems. (Full disclosure: I serve as a senior research fellow at the center.)

Dr. Boyd focuses on, in her words, “helping everyone else understand AI and its impact without the jargon, hype, or hot takes.” She specializes in helping “mission-driven” organizations use AI in ways that match their values. In other words, if you’re looking to use AI for good and not evil, Dr. Boyd can help.

The Ecological Impact of AI

When asked about the environmental impacts of AI, Dr. Boyd agrees that it’s “a really good thing to be concerned about. I think we definitely should be paying attention to the environmental impacts of technology, and I love that AI is kicking off that conversation.” 

“But,” she adds, “this is a data-center problem, not an AI problem. Streaming video for video calls, Netflix, and YouTube are all more resource intensive than AI is.” 

Data centers are the computing hubs that power all the computer-enabled magic of modern life: AI, cloud computing, video streaming, social media, and everything else.

Dr. Boyd notes that popular estimates like “AI uses one bottle of water for every query” or “a ChatGPT question uses 10 times as much energy as a Google search” are inaccurate and that water and energy use of other common activities (like your morning cup of coffee or streaming an episode of The Pitt) actually dwarf that of AI queries. 

She suggests that a person could offset their annual AI-related water use by avoiding one hamburger. For an organization, she says, “giving staff one extra day to work from home per quarter would way more than cover their LLM [large language model] use.”

Greener Pastures for AI

That said, for Dr. Boyd, the focus on personal AI use can distract from the broader systemic changes that could substantially reduce the impact of data centers. Good policies are needed, she says, to encourage tech companies to use greener energy and pay for the costs that data centers’ large energy demands will have on our energy systems. 

Specifically, Dr. Boyd says, we need policies to site data centers carefully, make them transparent about their water use, replace dirty with sustainable energy sources, and protect the environment from the effects of mining the rare-earth minerals that are present in all electronics.

Local governments are starting to make strides, she notes, even if the federal government remains uninterested. And consumer pressure is starting to have an effect on tech companies.

It’s true that questions about sustainability from one customer, even a large one, may not motivate a tech giant, she says. “But the more these companies are aware that people are going to ask about that and that we are considering that in our acquisition decisions, the more likely they are to make changes.”

Large organizations, Dr. Boyd says, likely have more leverage because they're potentially “a really big customer for an AI company.” Individuals, who don’t have as much of a voice, can join organizations that advocate for systemic change. She mentions that the Climate Action Campaign, the Union of Concerned Scientists, and the Green Screen Coalition are working on data center impacts, and that the NAACP has a Stop Dirty Data Centers project.

The Upside of AI

Many AI apologists (Bill Gates among them) claim that AI will solve its own energy problem by developing more efficient computers and optimizing the grid. “I don't have any particular reason to doubt them, except that I am a pessimist,” Dr. Boyd says. “I do think there will be a lot of beneficial applications of AI that are really well worth the resource use, things like drug discovery and personalized medicine and accessibility.

“And so if indeed AI is going to have this huge impact on the world to the extent that it is positive, which is debatable,” she adds. “it would be really valuable for us to figure out how to make data centers less impactful on the environment and their communities.”

While we wait to see whether AI solves all our problems or makes them worse, there are ways in which AI can help right now, specifically in helping people and organizations minimize their own ecological footprints. A detailed environmental audit can be expensive, but AI can perform a simple one.

“Giving [AI] as much information as you can about your activities and finding places where you're being inefficient can really help,” Dr. Boyd says. “I think it's easy to focus on things like technology use. What about travel? Commuting is very often an enormous chunk, and people don't often think about environmental impact when we're thinking about things like work from home.”

Threats and Opportunities

Dr. Boyd quickly lists other ways in which AI could be problematic: job displacement; “de-skilling,” where people rely too much on AI; and the use of AI in law enforcement and criminal justice, which, she says, “just seems massively premature and very scary.”

But for her the question is not “to AI or not to AI.” 

“I think we can find a lot of ways to do more good in the world rather than just reducing the harm that AI can produce,” she says. “And I think that takes some creativity and critical thinking, but also facing our values concerns, our values conflicts, directly, rather than thinking, ‘Well, I have to not use it or I must use it.’”

You can order Dr. Karen Boyd’s book Amplify Good Work: Effective, Ethical AI for Mission-Driven Work here. 

Five Takeaways

• Data centers for all sorts of computing applications, not just AI, use significant amounts of electricity and water.
• But your personal or work-related AI use is not a large contributor to the problem and is easily offset by other, minor actions.
• To really reduce the environmental impact of data centers, we need policies that will make tech companies accountable for their environmental impact.
• Large customers and even individuals can pressure tech companies to be more resource-efficient.
• AI also offers opportunities to get greener. You can ask AI for the best ways to reduce the environmental impact of your organization or even your daily activities.

What You Can Do

If you’re part of an organization, use your leverage to ask your tech providers about their energy and water use. As an individual, you can support an advocacy group, like  the Climate Action Campaign, the Union of Concerned Scientists, the Green Screen Coalition, or the NAACP’s Stop Dirty Data Centers project. You can also use AI to do a quick environmental audit of your organization or personal activities, and it will give you ways to reduce your environmental impact.