Shaping AI

2021-2025 | Dominique Cardon, Maxime Crépel, Valentin Goujon, Pauline Gourlet, Donato Ricci | médialab (Sciences Po, France)

This international research project aims to compare the controversies and public issues surrounding artificial intelligence (AI) in four countries: Canada, Germany, the UK and France. In France, the focus was on civic participation, using design methods developed at Sciences Po médialab over the past 10 years. The goal was to go beyond the dominant framings on AI, be it technophobic or enthusiastic, that they identified from an analysis of the recurring themes in the French press, by building an alternative way of participating in AI’s problematisation. This participatory approach involved about 30 people who realised AI through their situated activities in a year-long collective inquiry. They are referred to as “AI practitioners”. The team broadened the notion of “AI practices” to include other activities that nonetheless participate in AI’s development. Therefore, AI practitioners participating in the co-inquiry were not only data scientists or AI researchers but also journalists, lawyers, public officials, artists, SHS researchers or activists. The co-inquirers were first met individually to talk about their activities and experiences with AI. Based on their testimonies, a corpus of over 1000 documents was compiled. During a second interview, they were invited to analyse this documentation, allowing them to structure their concerns in relation to their situated activities. This was video-recorded and used in a qualitative analysis from which 19 themes emerged. A two-day group workshop gathering all the co-inquirers was then held to explore and discuss these themes in greater depth, problematising AI from the plurality of their perspectives. This collaborative work involves researchers and practitioners. It is a long-term project, and it took over a year to structure, analyse and disseminate the material produced. The research team is now exploring educational extensions in collaboration with museums and cultural institutions. It is considering creating a collective to better represent civil society in public debates on AI. The project highlights a lack of competent intermediaries in France on these subjects and proposes to remedy this through training and co-construction.