Beyond urban metrics

In AI and Language in the Urban Context, I make the case that cities are substantially linguistic entities, their social, cultural, and material dimensions shaped and sustained through conversations. As large language models (LLMs) exert increasing influence within public life, they not only automate services but contribute to these urban conversations. The AI Index Report 2025 that I referenced in my last post offers an empirical snapshot of AI’s rapid diffusion—confirming, and occasionally complicating, the book’s central claims.

The report confirms a key premise of the book: the ubiquity of conversational AI in everyday urban systems. For instance, it notes that “78% of organisations now report adopting LLMs for key functions,” including education, customer service, and document generation. These uses echo the examples I provide in the book — such as ChatGPT’s ability to respond to prompts about urban graffiti as semiotic disruption, or to participate in the co-construction of narratives among professionals, citizens, students and researchers. The book presents such exchanges not merely as functional outputs but as contributions to a broader semiotic ecology within the city.

Public sentiment, too, reflects the mixed character of urban discourse. The AI Index Report reveals rising optimism about AI in most countries, even those previously skeptical, alongside persistent concerns over misinformation, bias, and data misuse. These ambivalences mirror what I describe as the “urban chatter” that pervades city life — a blend of idle talk, gossip, critique, and planning discourse that, while messy, constitutes a vital part of the urban fabric.

The report also highlights the growing use of LLMs for narrative and creative work. For instance, AI is increasingly deployed in generating ideas and for other design-oriented tasks, resonating with the book’s treatment of planners and designers as narrative agents. In Chapter 10, I draw on the work of James Throgmorton and Lieven Ameel to argue that city planning is a form of persuasive storytelling. LLMs, trained on expansive corpora of human discourse, extend this narrative mode.

My opening critique of the dominance of cybernetic models in smart city discourse contrasts with the absence of semiotic or humanistic framings in the AI Index Report. The report’s emphasis on benchmarks, efficiency, and investment signals a calculative logic rather than an interpretive one—effectively confirming the relevance of my alternative framing.

I am calling on ChatGPT for assistance with this post. It put the observation about my semiotic framing more pointedly than I could.

Notably absent from the AI Index Report are discussions of meaning-making, symbolism, or the built environment. While the report excels in charting technical benchmarks, investment flows, and geopolitical dynamics, it largely bypasses the interpretive and cultural questions at the heart of my book. It offers little that engages directly with architecture, urbanism, or the humanities more broadly. This omission underscores the need for perspectives that treat language not just as data, but as a constitutive force in urban life.

If LLMs are increasingly embedded in the tools, scripts, and rituals of urban interaction — as my book contends — then documents like the AI Index Report represent only part of the picture. What remains to be explored, and contested, is how these language models, whether embraced or resisted, reshape the narratives, spaces, and imaginaries of the city.

In the next post I’ll emphasise how AI technologies reveal and conceal aspects of urban living.

References

• Ameel, L., The Narrative Turn in Urban Planning: Plotting the Helsinki Waterfront, London: Routledge, 2020.
• Coyne, R., AI and Language in the Urban Context: Conversational Artificial Intelligence in Cities, London: Routledge, 2025.
• Maslej, N., L. Fattorini, R. Perrault, Y. Gil, V. Parli, N. Kariuki, E. Capstick, A. Reuel, E. Brynjolfsson, J. Etchemendy, K. Ligett, T. Lyons, J. Manyika, J.C. Niebles, Y. Shoham, R. Wald, T. Walsh, A. Hamrah, L. Santarlasci, J. Betts Lotufo, A. Rome, A. Shi and O. Sukrut, The AI Index 2025 Annual Report, Stanford CA: AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, 2025.
• Throgmorton, J., Planning as Persuasive Storytelling, Chicago: University of Chicago Press, 1996.

Note

ChatGPT generated the featured image in response to the prompt: “Please generate an image of a city made up of folded newspapers, but in disarray.”