Imperfectum Prototyping and Democratic AI Design: beyond technocentric solutions

On 13 January 2025, the UK Labour party unveiled a new AI plan aimed at positioning the UK as a glob- al leader in technology, vowing to “mainline AI into the nation’s veins” to revolutionise and rescue the country’s weakened public services (UK Government, 2024). This reflects a broader trend in political jus- tifications for AI’s integration into democratic processes, often shaped by AI epochalism (Leslie, 2022): a technocentric, presentist perspective that portrays AI as uniquely transformative while overlooking its socio-historical contexts. However, a tradition of research started with big data and platforms (Gillespie et al., 2020), has demonstrated that these political views often ignore the inherent uncertainty, inequity and unaccountability of AI-driven governance (Eubanks, 2018; Dencik et al. 2018). This paper introduces the concept of imperfectum prototyping as an alternative approach to AI integration in society, borrowed from design approaches (Gobbo et al. forthcoming). Imperfectum prototyping argues that AI integration in democratic settings should be treated as an evolving, contestable process rather than a fixed technical solution. It builds on participatory design traditions (Delgado, 2023; Gourlet et al., 2024) to conceptualise AI not as a finished system but as an iterative, open-ended democratic experiment (Gobbo et al. forthcoming; Marres et al. 2024). Through participatory design methodologies, we are interested in co-creating AI governance frameworks where citizens, policymakers, and technologists can collectively shape the governance of AI systems in ways that remain open to revision and contestation. By foreground- ing democratic deliberation, pluralism, and reflexivity in the adoption and implementation of AI systems in democratic societies, this contribution aims to stimulate a discussion on concepts that counteract top-down, technocentric approaches to AI governance models.