ABSTRACT
Algorithms are deployed in all the interstices of our individual and collective activities. They are transforming our daily lives without the question of their social acceptability being raised in the legal field. Little by little, lawyers and regulators are shaping the legal framework of AI. More precisely, regulators, especially in Europe, considered AI systems from a business approach. The European Commission proposes a regulation for AI, with a risk-based approach. The goal of the current legal framework is to assess the risks generated by an AI system and then to process guidelines to prevent the risks, even by banning an AI system if the risk is considered too high. This approach is top-down: the regulation is laying down on the relationship between AI businesses and public regulators. But one important dimension is unseen: that of the social acceptability of those systems. In other words, the involvement of consumers, individuals, and citizens is totally ignored. They are kept out of the discussion about the relevance and adequacy of AI for the public good. This article aims to demonstrate the need to include citizens in the debate around AI, to ensure that AI is genuinely made with people, in accordance with what the concept of human-centered AI implies.
