Research suggests that pollution caused by AI could cause 600,000 asthma cases and 1,300 premature deaths in the United States

A major study found that by 2030, the electricity needed to support artificial intelligence (AI) could cause serious public health problems in the United States, resulting in an estimated 600,000 cases of asthma and a public health burden of up to $20 billion. The study pointed out that with the rapid development of AI technology, data centers are consuming more energy to meet their intensive computing needs. This trend has led to a significant increase in emissions of air pollutants, which has had adverse effects on public health.

It is predicted that by 2030, these pollutants will cause approximately 600,000 people in the United States to suffer from asthma and may cause 1,300 premature deaths.

“Public health issues have a direct and far-reaching impact on people, far beyond the geographic area where data centers operate,” said Professor Shaolei Ren of the University of California, Riverside. “Because air pollution is contagious, rising levels of pollutants will affect the health of people across the United States.”

It is understood that in the United States, part of AI's electricity demand relies on burning fossil fuels, and the air pollutants produced by fossil fuel combustion, such as fine particulate matter, pose a serious threat to human health. Researchers gave an example that the air pollutants produced by the electricity required to train a large AI model are equivalent to the total amount of pollutants produced by a car driving more than 10,000 round trips between Los Angeles and New York.

To more accurately assess the health impacts of air pollution and emissions in the United States, researchers used tools provided by the U.S. Environmental Protection Agency to calculate. The results show that by 2030, the total public health costs of data centers across the United States could exceed $20 billion. This figure is about twice the public health costs of the U.S. steel industry, and may even be comparable to the health impacts of pollutants emitted by tens of millions of cars in several of the largest U.S. states, such as California.

In addition, the researchers also believe that the use of backup gas generators in the "data center cluster" in Virginia, the United States, may have had a serious impact on the health of local residents. It is estimated that these generators may have caused asthma symptoms in 14,000 local people, consuming about $220 million to $300 million in public health costs each year.

If the generators were used frequently and emissions reached the maximum level allowed by the state, the total public health costs could be as high as $3 billion per year. The impact would not only be limited to Virginia, but also spread to areas as far away as Florida.

Faced with this grim situation, some technology companies are actively seeking solutions. They are supporting low-emission energy projects, financing renewable energy, and investing in new nuclear reactor technologies for traditional nuclear power plants.

However, for now, many data centers still rely heavily on fossil fuels such as natural gas for electricity generation. Research shows that by 2030, data centers could increase U.S. natural gas demand by an amount equivalent to that of New York or California.

In this regard, Benjamin Lee of the University of Pennsylvania pointed out: "The impact of AI and data center computing on health is indeed an important issue. However, the basic approximations and assumptions behind the specific numbers need to be verified through additional research."

Zikuai Technology