AI adoption in public sector urban planning: Aspirations and realities in London and Beijing

This paper examines the adoption of artificial intelligence (AI) in public sector urban planning through a comparative analysis of London and Beijing. Drawing on rapid qualitative methods combining interviews with public sector planners and related actors with documentary analysis, it adapts and spatialises Wisdom et al.’s multi-level framework across four analytical dimensions: public policy contexts, financial conditions, legal frameworks and regulations, innovation, data and organisational determinants and individual planners’ values, skills and norms. A central finding is that AI means different things to different planners. Interviewees drew on two broad understandings: general-purpose large language model tools used informally for drafting and synthesis, and purpose-built planning-specific tools tied to particular statutory functions. In both cities, adoption is largely confined to the former, concentrated on downstream tasks such as drafting documents, while the latter remains exploratory and uneven. The paper demonstrates that barriers to adoption outweigh drivers, raising key questions about the uncertain direction of AI-led public sector planning delivery and its associated politics. Across both cases, the adoption of AI is less a narrative of inevitable technological transformation than a process of political, institutional and professional negotiation over what is considered acceptable, trustworthy and feasible. By geographically locating where AI is being developed, tested and used in planning and tracing how its influence is distributed across distinct urban governance contexts, the paper argues for repositioning debates on AI in planning away from deterministic narratives of technological transformation and toward a more critical, geographically and governance-led understanding of adoption and non-adoption.