This research focuses on the acceptability of smart replies in emails and text messages. We propose a set of communication scenarios (both in a professional or personal context) to users and ask them what their reply would be in each scenario, offering them a pool smart replies to chose from or compose their own message. Our goal is to determine whether users find it interesting or not to use these suggestions. Analysis of the user responses will allow us to identify scenarios eliciting more user acceptance toward the smart replies. Finally, analysis of their answers and qualitative interviews will highlight how they perceive these technologies and their challenges (sincerity, agency, etc.).

Controlling or supervising critical systems exposes operators to increasingly complex problems to manage and to an ever-growing mass of data. These problems constitute a major challenge for operators, particularly in terms of selecting relevant information, understanding the situation, and making decisions, especially in contexts of high time pressure. The advances made in Artificial Intelligence open new perspectives for assisting operators in these contexts. In this paper, we present the user-centered design approach developed to design an AI-based system whose objective is to assist operators in these contexts. This system is characterized by its genericity; it covers different work situations involving critical systems (airplane, information system, electrical network, railway transport). It aims at supporting operators in real time in piloting or supervision activities.

« Le progrès et la catastrophe sont l'avers et le revers d'une même médaille ». Cette maxime de la philosophe Hannah Arendt, s’applique particulièrement à l’Intelligence Artificielle (IA). Comme toute technologie disruptive, l’IA progresse sur une corde raide entre bénéfices et risques, sur les plans social, sociétal et environnemental. Formidable potentiel de progrès au service de l’humanité et de la planète, elle est également porteuse de risques inédits, qu’ils soient directs (pour les utilisatrices et utilisateurs) ou indirects (pour les individus non-utilisateurs, les relations entre individus, la société, l’environnement, etc.). Sa conception et son utilisation requièrent donc une vigilance également inédite. Dans ce contexte, il est important que les entreprises aguerrissent les capacités réflexives individuelles et collectives des collaborateurs et collaboratrices dont les métiers consistent à concevoir, développer, exploiter et utiliser une solution d’IA. L’objectif est qu’ils et elles soient capables d’exercer leur responsabilité et d’aborder les questions éthiques soulevées.