Increased computational demands, enhanced data volumes, and cooling infrastructures have led to the rapid expansion of energy-intensive data centers. However, AI-powered applications and services are the key contributors to this trend.
This increased power requirement for Artificial Intelligence is spurring increased levels of deadly air pollution from generators and power plants. This could impact the environment through escalating carbon emissions and water consumption.
“The growth of AI is driving an enormous increase in demand for data centers and energy, making it the fastest-growing sector for energy consumption across all industries,” says the co-author Shaolei Ren.
Findings by researchers at the University of California Riverside and Caltech have estimated around 1,300 premature deaths per year by 2030 in the United States. The total public health cost from cancers, asthma, other diseases, and missed work and school days are assessed to be $20 billion a year.
Still, the tech giants appear to neglect these human and financial costs. Though these tech industry reports mention carbon emissions and water, there’s hardly any comment on unhealthy air pollutants.
With the current competition in furnishing better AI services, the air pollution caused by the power plants that power AIs is resulting in lung-penetrating fine particles smaller than 2.5 micrometers. Other pollutants like nitrogen oxides are expected to increase steeply.
“If you look at those sustainability reports by tech companies, they only focus on carbon emissions, and some of them include water as well, but there’s absolutely no mention of unhealthful air pollutants and these pollutants are already creating a public health burden,” said Ren.
Scientists urge to resist reliance on “bathtub modeling” of flood risk
To illustrate better, the researchers calculated the emissions of Meta’s Large Language Model Llama-3.1. The pollution caused by training the LLM is equivalent to 10,000 car trips between Los Angeles and New York City.
The study also found that this air pollution affects certain low-income communities due to their proximity to power plants. This pollution then drifts into the country.
For instance, pollution from generators at data centers in Northern Virginia glides into Maryland, New York, or New Jersey, costing $190 million to $260 million a year in public health costs. Shockingly, these generators can even become 10-fold and reach $1.9 billion to $2.6 billion.
Even in some areas, the public cost exceeds the pay by tech companies for electricity.
“The data centers pay local property taxes to the county where they operate. But this health impact is not just limited to a small community. Actually, it travels across the whole country, so those other places are not compensated at all.“
The authors of this study recommend that tech companies adopt standards and methods for reporting the air pollution caused by their power plants. They also recommend that the victim communities be properly compensated for this health burden.
“If you have family members with asthma or other health conditions, the air pollution from these data centers could be affecting them right now. It’s a public health issue we need to address urgently,” Ren said.
Blacks have more exposure to air pollutants, increase heart disease risk
Journal Reference
- Han, Y., Wu, Z., Li, P., Wierman, A., & Ren, S. (2024). The Unpaid Toll: Quantifying the Public Health Impact of AI. ArXiv. DOI: abs/2412.06288