For a “situated” conception of artificial intelligence

The purpose of the 220-221 issue of the journal Réseaux that we coordinated on the design and uses of conversational agents was to clarify a sociological perspective on Artificial Intelligence, putting aside both promotional discourses and principled denunciations of the “grip” of these technologies on society. In this paper, which has been occasioned by a one-day workshop, we propose a brief detour through the reactions to that issue and the avenues and research questions that it opened. We first discuss the implications of a sociological perspective on the technologies and practices associated with the label “AI”, particularly in the so-called “observational” approaches. We then return to the achievements of the issue in the field of human-robot relations and discuss the new questions they raise. Finally, we show how the results of observational approaches resonate with the concepts of relationship, agencing or socio-technical configuration.