In his recent encyclical devoted to artificial intelligence, the Pope calls for collective discernment in the face of contemporary technological transformations. Artificial intelligence (AI), he writes, cannot be thought of as a simple question of innovation or competitiveness. It involves our way of living in the “common home”, of sharing resources and of guiding progress in the service of the common good.
This call comes at a time when an argument systematically comes up whenever we worry about the environmental cost of artificial intelligence: its benefits will ultimately offset its costs. Why be alarmed about the energy consumption of data centers if, tomorrow, AI will make it possible to optimize electricity and transport networks, buildings or certain industrial processes?
Environmental compensation: a sham argument
This “compensation” argument is, however, based on incomplete reasoning. It selects certain favorable uses and ignores others. However, the effects of applications of artificial intelligence are multiple. Some actually contribute to reducing the carbon footprint or better managing resources.
Others have no particular environmental objective, such as when a conversational model generates entertainment content or large-scale personal uses. Finally, others directly aggravate environmental pressures through the optimization of oil extraction or the acceleration of targeted advertising.
The problem is therefore less that of “AI” in general than that of collective arbitrations on its uses. The encyclical rightly recalls that no technology is neutral when it transforms social and economic structures. As Francis already wrote in Laudato si’, the risk of a “technocratic paradigm” appears when efficiency becomes the only decision criterion, without reflection on the goals pursued.
Understanding AI
But today, we do not even have the data necessary to carry out this democratic discernment. At the Observatory on the Environmental Impact of AI, a project of the IA and Society Institute of the École Normale Supérieure (ENS) scientifically directed by university professors Aurélie Bugeau and Anne-Laure Ligozat, we are confronted daily with this opacity. Researchers are investing a lot of time to develop innovative methods to obtain the information needed to understand the extremely rapid evolution of the AI industry.
The main blind spot is regulatory. The European AI Act imposes certain transparency obligations on AI models; The EU Energy Efficiency Directive requires information about the data centers themselves. But between the two, nothing: we don’t know what actually works inside these infrastructures.
How much is really AI? How much is spent on training the models, how much on using them? What uses consume the electricity and available computing capacities? How much is dedicated to public services or medical applications? This question, however, becomes central. The global race for AI is already creating tensions over access to energy, semiconductors, electronic components for computing, and digital infrastructure.
Discuss uses
These resources compete with other essential needs: electrification of transport, thermal renovation of buildings or industrial modernization. The Shift Project estimates that data centers could represent 7.5% of French electricity consumption in 2035, compared to around 2% today.
It is here that the call for discernment launched by the Pope finds a concrete political translation. It is neither a question of rejecting technological innovation nor of adhering to naive enthusiasm. But to be able to debate uses and distinguish those which truly serve the common good from those which aggravate ecological and social imbalances.
France, which in 2025 became the first European country to host foreign investments in data centers thanks to its carbon-free electricity, has significant leverage. It could condition this attractiveness on a requirement for transparency on uses and digital infrastructures, or even set priorities in favor of services meeting national interests.
Without this transparency, the compensation argument will remain an unverifiable promise, mobilized to close the debate rather than to enlighten it. Conversely, making transparency a central principle would perhaps make it possible to reconcile technological innovation and ecological responsibility, and to make AI a lever in the service of a fairer, more sober and more livable society.
About opinions
This text is signed by a guest author. He expresses his opinion and not that of the editorial staff. Our Live section aims to allow the expression of pluralism on religious, social and current affairs subjects, and to encourage dialogue, according to the criteria set by our editorial charter.
Share your opinion in comments or by writing to us at: readers.lacroix@groupebayard.com
