A groundbreaking study from researchers at the University of California, Riverside, and the California Institute of Technology has revealed the staggering public health costs of artificial intelligence’s environmental impact. The research, titled “The Unpaid Toll: Quantifying the Public Health Impact of AI,” estimates that AI-related air pollution could cost $20 billion in health impacts by 2030, representing what may be the first comprehensive assessment of AI’s contribution to air quality degradation.
The study found that training a single large language model like Meta’s Llama 3.1 generates as much air pollution as a car driving round trip from New York to Los Angeles 10,000 times. By 2030, electricity generation for AI data centers could lead to 1,300 additional premature deaths annually—a 36% increase over current asthma-related deaths in the United States. The researchers also project approximately 600,000 asthma-symptom cases per year by the end of the decade.
Led by Shaolei Ren of UC Riverside and Adam Wierman of Caltech, the research examined emissions of nitrogen dioxide, sulfur dioxide, and fine particulate matter (PM2.5) from power plants and diesel generators serving AI facilities. The AI boom has dramatically increased electricity demand, with McKinsey & Co. projecting data centers will consume 11-12% of total US electricity by 2030, up from just 3-4% last year.
The study revealed particularly alarming findings about diesel backup generators at data centers, especially in Virginia, home to one of the world’s densest data center concentrations. These generators produce 200 to 600 times more nitrogen dioxide per unit of power than natural gas plants. Even at just 10% of permitted emissions, Virginia’s data center generators could cause 13-19 additional deaths annually, with public health costs reaching $220-300 million per year.
Crucially, the research found that air pollution from AI facilities crosses state lines, with Maryland’s Montgomery County most affected by Virginia’s AI generators. Effects extend to West Virginia, New York, New Jersey, Pennsylvania, Delaware, Washington DC, and even Florida. The study also highlighted that economically disadvantaged communities bear a disproportionate burden of these health impacts. Last year alone, the generative AI boom created an estimated $5.6 billion public health burden.
Key Quotes
There is something like this, air pollution, which is affecting people right now. We aren’t paying attention to it at all.
Shaolei Ren of UC Riverside emphasized the immediate and overlooked nature of AI’s air pollution impact, highlighting how the industry has focused on carbon and water usage while ignoring direct health consequences affecting communities today.
Diesel generators represent a major source of on-site air pollutants for data centers and pose a significant health risk to the public.
The researchers identified backup diesel generators as a particularly deadly component of AI infrastructure, producing 200 to 600 times more nitrogen dioxide than natural gas plants and potentially causing up to 190 additional deaths annually in Virginia alone.
We thought the air pollution was limited to a small area. That’s not true. There is actually cross-state air pollution.
Ren explained how their research revealed the far-reaching geographic impact of AI-related air pollution, with Virginia’s data centers most severely affecting Maryland’s Montgomery County and extending effects as far as Florida.
If you look at the sustainability reports from these companies, they mention carbon and water, but they don’t mention anything about air pollution. They should start reporting this in the same way.
Ren called out major tech companies including Amazon, Google, Microsoft, and Meta for failing to disclose air pollution impacts in their annual sustainability reports, advocating for greater corporate transparency on this critical health issue.
Our Take
This study fundamentally challenges the narrative that AI’s primary environmental concerns are limited to carbon footprints and water consumption. The $20 billion health cost projection by 2030 represents a hidden subsidy that communities—particularly economically disadvantaged ones—are paying for AI advancement. The cross-state pollution findings are especially significant, as they complicate regulatory approaches and reveal how AI infrastructure decisions in one location create health burdens elsewhere. The diesel generator findings are particularly troubling, suggesting that backup power systems designed for reliability are creating ongoing health crises. This research should prompt immediate action from both regulators and AI companies to implement cleaner energy sources, optimize training schedules for lower-impact periods, and fundamentally reconsider data center locations. The lack of transparency from major tech companies on air pollution impacts suggests either ignorance or willful neglect of a critical externality that society is bearing.
Why This Matters
This research represents a critical turning point in understanding AI’s true environmental and societal costs. While the tech industry has faced increasing scrutiny over carbon emissions and water usage, the direct health impacts of air pollution from AI infrastructure have been largely overlooked. The findings reveal that AI’s public health costs could exceed those of coal-based US steelmaking and rival emissions from California’s 35 million cars by 2030.
The study exposes a significant gap in corporate transparency, noting that major tech companies like Amazon, Google, Microsoft, and Meta fail to report air pollution impacts in their sustainability reports. This oversight is particularly concerning given the disproportionate impact on economically disadvantaged communities and the cross-state nature of the pollution. The research could catalyze regulatory action and force AI companies to reconsider data center locations, training schedules, and backup power sources. As AI adoption accelerates across industries, understanding and mitigating these health impacts becomes essential for sustainable development of the technology.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources:
Recommended Reading
Related Stories
- How AI Can Help Make Buildings More Energy Efficient
- EnCharge AI Secures $100M Series B to Revolutionize Energy-Efficient AI Chips
- The Artificial Intelligence Race: Rivalry Bathing the World in Data
- Wall Street Asks Big Tech: Will AI Ever Make Money?
Source: https://www.businessinsider.com/ai-20-billion-dollar-air-pollution-problem-2024-12