The existence of artificial intelligence (AI) stretches back decades, from bots that could play games in the 1950s to the virtual assistants incorporated into smartphones in the 2010s. But in November 2022, AI use expanded drastically after the company OpenAI released ChatGPT, a free generative AI chatbot, to the masses, allowing users to enter text or image prompts that the chatbot could process into instant responses.
These exchanges range from asking the bot to generate a meal plan to asking it to summarize scholarly research papers – with either task taking the bot mere seconds. Within days of its debut, the app surpassed 1 million users. (As of this writing, at least 100 million people use the platform daily.)
Following ChatGPT’s release, AI tools spread widely and rapidly as tech companies scrambled to incorporate AI features into their products. Proponents have lauded this technological breakthrough and what they see as AI’s potential to enhance the convenience of everyday tasks, improve worker productivity, advance scientific research, and more. Critics point to the risks of plagiarism, threats to privacy, the spread of misinformation, and the displacement of human workers and human problem-solving. (Some have even hypothesized about whether the tech might “go rogue” and harm humans.)
Both favorable and critical stances touch on the environment, with some saying AI could help with resource management and extreme-weather tracking, and others pointing to AI’s appetite for nonrenewable resources. Just as climate scientists can use AI to optimize renewable energy systems and find leaks in pipelines, the fossil fuel industry can use it to extract more oil and gas.
Thirst Trap
While AI may seem like it resides in an intangible “cloud,” the infrastructure it requires to run takes up a lot of physical space – and tangible resources. Sprawling, temperature-controlled data centers use copious electricity, as well as freshwater to cool the hardware. As AI models “learn” (modify their behavior with experience), their resource needs increase.
The U.S. has more than 5,000 data centers, thousands more than any other country, with 80 percent of U.S. data centers concentrated in 15 states, including California, Arizona, Texas, Illinois, and Virginia. These data centers collectively consume as much electricity as entire nations. According to MIT News, by 2026, U.S. data centers will rank fifth on the list of largest electricity consumers in the world, between Japan and Russia. By 2030, International Energy Agency analysts predict data centers will increase U.S. electricity demand by 20 percent. This explosive growth is happening at a time when many countries have committed to cutting carbon emissions and reducing their reliance on nonrenewable energy sources.
As for water, a single data center can consume 18,000 gallons of water per day, the equivalent of 4,200 humans’ daily needs, at a time when the United Nations estimates that nearly half of the world’s population – 4 billion people – experiences severe water shortages for at least one month a year. These resource requirements strain energy grids and water supplies, from diverting freshwater in Arizona, which is among the states that recently faced the Southwest’s worst drought in a millennium, to contributing to rate hikes in New Jersey, where the state’s public utilities board warned residents they could see 20 percent increases on their electricity bills. In Virginia, the National Parks Conservation Association found that the construction of two proposed data centers could result in 57,000 tons of sediment being dumped into the Occoquan Reservoir, which supplies drinking water to more than 800,000 people.
Tech’s Wild West
Utah and Colorado are among a few states that have put forth legislation on AI to protect consumers, but they’re up against the powerful tech industry, hungry for a share of a market poised to hit nearly $5 trillion by the 2030s. Consumer advocacy groups are pushing for stronger oversight, with a coalition of 90 consumer rights and environmental justice groups creating a People’s AI Action Plan to promote a “trajectory for AI that puts people first, rather than the interests of tech billionaires.”
“The frenetic pace of development [is] akin to a Klondike gold rush,” write researchers at the Massachusetts Institute of Technology, who advocate for “a more deliberate, thoughtful approach that considers long-term environmental impacts and societal needs.” So far, no federal legislation exists to regulate the development or use of AI, leaving consumers to their own devices when weighing convenience against the costs. And considering that producing a single AI-generated image can use the equivalent of a bottle of water or as much energy as charging a smartphone, the costs are high.
