Duarte, Magalie <br/><br/>
Understanding the complexity of an artificial intelligence adoption situation in the public sector, by mobilizing Edgar Morin’s dialogical principle 

 -  2025.<br/>							 
<br/><br/> 89 
<br/><br/> The present study explores the process of adopting artificial intelligence (AI) in the public sector through the lens of complex thinking and, especially, Edgar Morin’s dialogic principle. The use of public AI creates significant societal and organizational tensions. Existing literature characterizes these tensions and provides insights into the rationale (the what and the why) behind AI adoption. The present study focuses on the process of AI adoption itself. The dialogic principle is mobilized, seeking to unite complementary and antagonistic movements rather than overcome them. It argues that this principle is essential to describe and understand these movements during the decision-making process to adopt AI. The present study adopts a qualitative and interpretive methodology, based on six months of immersion in the innovation department of a social housing provider, working on a predictive model of residents’ risk of loss of autonomy. The adoption process is characterized by the emergence of three antagonisms, as highlighted through our narrative approach: firstly, the tension between algorithmic rationalization and the mission of social support; secondly, the opposition between tacit and explicit knowledge within community-based professions; and third, the dilemma between surveillance and privacy. The dialogic reading proposed here, aligned with Morin’s intricate thought processes, helps facilitate a deeper understanding and resolution of these antagonisms compared to a binary or factorial decision model of whether or not to adopt AI. The study thus proposes a renewed analytical framework to support public decision-makers in the process of adopting AI, emphasizing the importance of an integrative approach to technical, social, and ethical issues. This study provides insight into the process of adopting AI in the public sector and encourages in-depth reflection on the responsible and contextualized adoption of these technologies.