L'invité du mois David BOL, Professeur à UCLouvain - "The growing carbon footprint of generative AI services" - #50ansIRISA - IRISA

À l’occasion de ses 50 ans, l’IRISA accueille tout au long de l’année 2025 des experts en recherche informatique dans le cadre de sa série "Les invités du mois".
Le lundi 30 juin, David BOL, Professeur à l'UCLouvain donnera une conférence intitulée "The growing carbon footprint of generative AI services"

Ouvert à la communauté scientifique, aux étudiant.e.s et toutes personnes intéressée et ayant un minimum de connaissances dans le domaine
Inscription obligatoire - (conférence en anglais)
---------------------------------

The growing carbon footprint of generative AI services

Over the last two decades, we have acknowledged the rise of AI algorithms starting from recommendation systems to speech and face recognition to the recent explosion of generative AI services for question answering, text writing as well as image and music generation.In parallel, most of Humanity has realized the magnitude of the environmental catastrophe it has caused, with climate change and five other of the nine planetary boundaries that are transgressed. These transgressions have led Humanity in the Anthropocene, i.e. out of the safe operating space, where we clearly see the effects of climate change, biodiversity loss and pollution.
AI services are part of the Information and Communication Technologies (ICTs) that contribute to the exceedance of these boundaries. In particular, the ICT sector was responsible of 2-4% of the global greenhouse gas (GHG) emissions in 2020. Since then, the GHG emissions (or carbon footprint) of the ICT sector has increased significantly with key datacenter operators such as Google and Microsoft reporting that the fast growth of the AI processing capacity has prevented them from reaching their climate targets. Indeed, their GHG emissions increase despite their investment in renewable electricity purchase. Therefore, voices across the world advocate for a responsible use of AI services for a few reasons including environmental ones.
In this talk, we will analyze the global warming contribution of the ICT sector and the impact that AI service emergence have on them. We will see why efficiency improvement is not enough to reach net-zero climate targets for the ICT sector, which calls for complementing it with a sufficiency mindset and alternative business models.