The World Health Organization’s 2024 report recently called on world leaders to “reduce pollution-related diseases and reduce carbon emissions”, noting that air pollution – particularly the extremely small particles known as PM2.5 – cause about 7 million deaths per year. The data has not improved since then, although, fortunately, AI tools have improved greatly and offer new possibilities.
However, it is essential to recognize the links between regulating air pollution and mitigating greenhouse gas emissions to prioritize health. Regulating air pollution is one of the most effective ways to save lives, and as an added benefit, it also limits greenhouse gas emissions.
On February 7, 2024, the EPA strengthened the National Ambient Air Quality Standards for Particulate Matter to protect millions of Americans from harmful and costly health impacts like heart attacks and premature death. The agency has just reduced the annual primary (health-based) PM2.5 standard from 12 to 9.0 micrograms per cubic meter.
These stronger PM2.5 NAAQS will advance environmental justice by reducing particulate pollution, which disproportionately burdens communities of color and other vulnerable communities. According to EPA’s regulatory impact analysis, the regulations on the books and available control measures could reduce particulate pollution, leading to large net public health benefits of up to $46 billion (in 2032). WHO guidelines state that annual average concentrations of PM2.5 should not exceed five µg/m3. Some remote areas or places with excellent air quality can have levels as low as 1-3 µg/m³
Many sources of air pollution also produce greenhouse gases: coal-fired power plants, road traffic (especially diesel vehicles), energy use in buildings (including cooking and heating with wood and coal), unsafe waste disposal and open burning, and industry (including fossil fuel heavy machinery and brick kilns). Thus, if we address the sources of air pollution, we will make immediate progress in regulating GHG emissions.
Regulation of air pollution and greenhouse gases helps to grow the economy
Cleaner air leads to fewer pollution-related illnesses and premature deaths, which translates into significant economic benefits. A 2011 EPA study found that the Clean Air Act Amendments of 1990 are expected to bring $2 trillion in direct benefits to Americans by 2020, exceeding costs by a factor of more than 30 to 1. Globally , air pollution causes the loss of 1.2 billion working days per year, which could reach 3.8 billion days by the year 2060, according to the non-profit Clean Air Fund. The evidence clearly shows that prioritizing clean air is consistent with economic development and can foster it by creating healthier, more productive populations and stimulating technological advances. According to the EPA, between 1970 and 2019, aggregate emissions of common air pollutants in the US fell 77%, while GDP grew 285%.
How AI tools can help reduce pollution
We now have the data and AI tools to inform data-driven decision-making based on facts and data, not opinion. Despite the heated debate over whether humans cause climate change, one reality is undeniable: exposure to air pollution and climate-related stressors, including extreme heat, wildfires, droughts and tropical cyclones, kills people, causes harmful health effects and causes disasters. consequences in society and economy,
However, thanks to the technological revolution, our ability to access massive amounts of data and the explosion of human-centric AI, we have the tools to make rapid progress in regulating pollution, mitigating greenhouse gas emissions and providing strategies based on US and global climate adaptation data.
AI algorithms can now simulate and improve climate models, reducing computation time and increasing the reliability of climate forecasts. AI tools can even be effectively deployed to assess air pollution exposure at a very granular spatial level. We can jointly model exposure to air pollution and greenhouse gas emissions and policy interventions and determine the most effective regulatory actions that will have the most benefit in reducing both while preserving the economy. We can deploy human-centered artificial intelligence models that will allow us to determine when, where, and how specific individuals and populations will be most affected by climate stressors and inform how best to intervene.
How AI tools can increase air pollution
All these approaches will save lives. However, AI plays a complex role in climate change mitigation efforts, with both positive and potentially worrisome impacts. Considering the future perspective of AI, we must assume that while AI offers significant potential to accelerate climate action, there are also valid concerns about its environmental footprint. According to the International Energy Agency, 2022 global data center electricity consumption accounts for 1-1.3% of global final electricity demand. A recent article in the MIT Technology Review cites a Harvard study – which I co-authored – where we reported that since 2018, carbon emissions from US data centers have tripled. For the 12 months ending August 2024, data centers were responsible for 105 million metric tons of CO2, accounting for 2.18% of national emissions (by comparison, domestic commercial airlines are responsible for about 131 million metric tons). About 4.59% of all energy used in the US goes to data centers, a figure that has doubled since 2018.
In summary, while AI offers powerful tools for climate action, its environmental footprint needs careful management. We must measure and monitor the responsible development and deployment of AI technologies and ongoing efforts to improve their energy efficiency and sustainability. There are many potential strategies to limit the carbon footprint of AI (for example, developing and using more efficient AI algorithms or increasing the energy efficiency of data centers using free air cooling). However, these approaches are in their infancy and their effectiveness has not been studied. We must balance the pace of AI penetration in every sector of society while considering its potentially harmful social consequences.
