AI’s climate problem is worse than we thought

A recent peer-reviewed study published in npj Climate Action reveals that the climate impact of artificial intelligence extends far beyond the energy consumption of its data centers. Former Microsoft employees Holly and Will Alpine, alongside other researchers, developed a global economic model to evaluate what they term "enabled emissions"—the carbon pollution made possible when tech companies provide custom AI tools to help fossil fuel corporations discover and extract oil and gas more efficiently. The findings indicate that Big Oil's use of AI could generate between 3.3 and 13.3 times more climate pollution than the electricity powering AI data centers, potentially matching the annual emissions of entire nations like Mexico or Russia. Furthermore, running simulations where AI benefits both renewable energy and fossil fuels simultaneously showed that emissions consistently rise unless AI aids renewables at an exponentially higher rate, or tech companies completely halt the creation of production-boosting tools for the fossil fuel industry. The researchers argue that public discourse and regulatory frameworks must expand beyond data center energy use to address and govern the downstream emissions enabled by technology.

(Source：Heated World)