AI's Climate Impact Goes Beyond Data Center Power Use, Scientists Warn

According to a study by American scientists, the climate impact of artificial intelligence is not limited to the electricity consumption of data centers. AI is increasingly used by major oil companies to find new deposits, drill wells faster, and extract more oil, thereby boosting fossil fuel production and harmful emissions. The researchers gathered data on AI's potential to improve efficiency in both fossil fuel and renewable energy sectors, then combined this with a model of the global economy to track the spread of effects. They simulated 64 scenarios with varying degrees of AI-driven productivity gains. Whenever AI had an equal impact on both sectors, global emissions in the model increased. For renewable energy to offset the harmful impact of fossil fuels, it would need five times more AI support. The only scenarios that reduced emissions were those where tech companies completely ceased helping fossil fuel extractors. The calculations are estimates, not actual measurements, but they suggest that under an optimistic scenario, AI's impact on global emissions is 3.3 times greater than data center-related emissions, equivalent to Mexico's emissions; under a pessimistic scenario, it's 13.3 times, equivalent to Russia's emissions. The scientists urge broadening the public discussion to include emissions from AI-optimized resource extraction, though they emphasize they are not against AI per se.