A Climate Fix? Promotional Discourse on Artificial Intelligence and the Environment Before and After ChatGPT
Théophile Lenoir¹,², Andrei Mogoutov²
DOI : 10.65701/67cukrb3xp
Corresponding author:
theophile.lenoir@sciencespo.fr
Abstract
This paper analyses a corpus of 1,561 press releases published between 2019 and 2025, examining how organisations frame AI as a response to environmental problems before and after the emergence of generative AI in public debates in November 2022. We code each press release for AI technique, problem, proposed solution, and sector. We report three findings. First, the vocabulary used to describe AI became less specific over time, with a growing share of press releases naming no specific technique. Second, over the period, AI’s own energy and environmental footprint, as well as energy production and grid stress, displaced climate change and industrial emissions as dominant problems. In response, solutions shifted from narrowly technical fixes associated with software capabilities (detection, forecasting, optimisation) toward hardware improvements, reporting, policy and investment. Third, the ICT sector overtook all other sectors as the largest producer of environmental press releases, far above sectors traditionally associated with AI’s environmental applications such as transport, agriculture and industry. After November 2022, the canonical use cases of AI for the environment are thus proportionally less prominent in promotional discourse, and AI’s own environmental footprint has become the central concern.
