A study has found that AI-driven productivity gains are likely to enable more planet-heating pollution from fossil fuels than they prevent from renewable energy sources. Researchers modelled 64 scenarios, concluding that net yearly carbon pollution could rise by 0.47-1.8 gigatonnes, representing approximately 1-5% of the energy sector’s annual emissions.
The research is the first to quantify AI's climate impact across the entire power sector. Previous studies had focused on indirect climate benefits, such as optimising electricity grids, without accounting for increased pollution from enhanced fossil fuel extraction.
According to the International Energy Agency, AI could increase technically recoverable oil and gas reserves by 5% and reduce the cost of deepwater offshore projects by 10%. Oil and gas executives have expressed enthusiasm for AI's role in finding additional fuel, with some describing its impact as "the next fracking boom."
The study's co-author, Holly Alpine, noted that the research used a conservative assumption that AI adoption would occur at similar rates for both fossil fuels and renewables. She highlighted that fossil fuel applications are already being deployed at scale, supported by industry evidence, while renewable applications remain largely in pilot or academic stages.