Charting Artificial Intelligence – CHAI

Artificial Intelligence (AI) is a highly polysemic and ever-evolving field, encompassing a diverse set of actors with heterogeneous positions and principles. This includes researchers from various fields, be it exact sciences or social sciences, regulatory bodies, companies involved in developing or utilizing AI models, media outlets, and associations critical of AI implementation in various social contexts, among others. It is indeed a construct with both social and technical dimensions that should be studied as such. Institutions such as the European Union Council, in its working document on the AI Act, emphasizes the importance of clarifying the roles of actors who can contribute to the development of AI systems.

We aim to understand how knowledge about AI is constructed and evolves. We propose to map this social landscape by examining the relationships between these actors, illustrating how the analysis of these relationships and their representation through enriched graphs (dynamic, multi-layered, etc.) allows for encoding the heterogeneity and dynamics of this social world. We will investigate how these graphs, by enabling the modeling of the semantic structure of texts addressing AI, can shed light on definitional and normative issues within the field. Lastly, we will qualitatively study the dynamics of collaboration and competition in relation to AI, especially by tracing the trajectories of researchers involved in AI and how these experts position themselves concerning media, scientific, and institutional discourses.

Thus, the questions surrounding graph modeling should help establish both the boundaries of the social world and its "topography", questioning the definitional boundaries of AI and how the trajectories of actors are influenced by, and in turn influence, this social world.