AI Could Boost Fossil Fuel Emissions, Study Warns

Artificial intelligence's application in the oil and gas industry could lead to a substantial increase in global energy emissions, according to new research published last week in the journal npj Climate Action. The study, authored by former Microsoft sustainability workers Will and Holly Alpine, cautions that AI's ability to enhance productivity in fossil fuel extraction may have a more damaging environmental impact than the emissions generated by powering AI data centers. The research indicates that the additional yearly emissions from AI-driven fossil fuel production could range from the equivalent of Mexico's emissions to as much as Russia's, the world's fourth-largest emitter. This projected increase in greenhouse gas emissions significantly outweighs the benefits AI offers to the development of solar, wind, and other clean energy technologies. Furthermore, the study finds that these emissions projections substantially exceed the anticipated emissions from the global buildout of data centers. Will Alpine, a co-author of the research, described the relationship between technology companies and the fossil fuel sector as a "self-reinforcing effect between supply and demand," emphasizing that these two aspects cannot be considered independently. He stated, "One of the key insights of our paper is that you cannot treat them independently. They are two sides of the same coin." The Alpines, who both formerly worked in sustainability roles at Microsoft, resigned from their positions at the beginning of 2024. Their departure was motivated by Microsoft's continued engagement with the oil and gas industry, prompting them to publicly advocate for greater awareness of the connection between AI and fossil fuels. The paper highlights that oil and gas companies have utilized various forms of AI for decades to improve the efficiency of locating and developing underground resources. The Alpines argue that this increased efficiency contributes to a greater global reliance on fossil fuels, thereby hindering efforts to achieve climate targets. While technology companies typically measure their own operational emissions and those within their supply chains, the research points out a lack of comprehensive measurement regarding the downstream emissions enabled by their AI technologies in sectors like fossil fuel extraction. The study's findings suggest a critical need to re-evaluate the net environmental impact of AI, particularly when its advancements facilitate the expansion of carbon-intensive industries.