Who Gets to Shape AI? Power, Creativity, and the Limits of GovernanceFriday 13 November 2026, 15:30 – 17:00

Mapping Public Value in AI Safety Research

  • Canhui LiuUCL Computer Science / UCL Centre for Artificial Intelligence

As AI attracts unprecedented investment and policy attention, the metascience tools used to evaluate emerging technologies risk overlooking the directionality of innovation — the public values and societal needs that research is positioned to address. Taking public research on AI safety as a case, this paper asks what public values are imagined in this literature and how societal needs are framed. We combine network analysis of bibliometric data and Topic Modelling of public discussions with grounded, interpretive analysis of framings in journal articles, complemented by expert interviews. The paper contributes empirical insights into the values and absences expressed in AI safety research, alongside a novel set of analytic techniques that integrate interpretive and network methods to fold 'the social world' into metascientific innovation mapping.

Keywords: directionality, AI safety, metascience, interpretive analysis

Panel

Who Gets to Shape AI? Power, Creativity, and the Limits of Governance

Friday 13 November 2026, 15:30 – 17:00

Chair: Kieren Reed — UCL, Slade School of Fine Art

As generative AI moves from experimentation to infrastructure, this panel examines how it is reshaping power, ownership and responsibility across creative and cultural production. Bringing together perspectives from speculative futures, intellectual property law, filmmaking and critical media studies, it considers how AI establishes new norms of creativity, legitimacy and value while complicating consent, authenticity and accountability. The discussion asks whether responsible AI requires adapting existing institutions or redistributing the power to shape technological change.