The Synthetic City: An Introduction

There is an emergent “structure of feeling”¹Raymond Williams, The Long Revolution (Parthian Books, 2014). among scholars, artists and practitioners that recent developments under the broad umbrella of “AI” herald potentially momentous shifts in how we live, work and play in cities. Urban environments have for some time been understood as venues for various kinds of computational agency, such as networked surveillance systems, platformised delivery apps, neighbourhood social media, automated street infrastructures or sensor-based advertising. Scholars have deployed a panoply of concepts to denote such phenomena, including “algorithmic cities,” “code/spaces,” “data-driven urbanism,” “platform urbanism,” “sentient cities,” “smart cities” and “urban digital twins.” Yet while “artificial intelligence” is hardly new – the notion has surpassed its 70^th anniversary²John McCarthy et al., “A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence, August 31, 1955”, AI Magazine, 2006, https://doi.org/10.1609/aimag.v27i4.1904. – the more recent release of easy-to-use, consumer-oriented generative AI tools such as ChatGPT, Claude, Microsoft Copilot, Stable Diffusion, Midjourney or Suno has spawned urgent and vigorous popular debate.³Kate Crawford, Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (Yale University Press, 2021); Emily M. Bender et al., “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?”, Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, 3 March 2021, 610–23, https://doi.org/10.1145/3442188.3445922; Murray Shanahan, “Talking about Large Language Models”, Communications of the ACM 67, no. 2 (2024): 68–79, https://doi.org/10.1145/3624724. In tandem, interdisciplinary academic debate has grown in a slightly different register, where the rise of “urban AI” marks the moment in which the computationally-enabled urban automation of smart cities is said to give way to increasingly autonomous, agency-wielding urban AI systems.⁴Aale Luusua et al., “Urban AI: Understanding the Emerging Role of Artificial Intelligence in Smart Cities”, AI & SOCIETY 38, no. 3 (2023): 1039–44, https://doi.org/10.1007/s00146-022-01537-5; Federico Caprotti et al., “Why Does Urban Artificial Intelligence (AI) Matter for Urban Studies? Developing Research Directions in Urban AI Research”, Urban Geography 45, no. 5 (2024): 883–94, https://doi.org/10.1080/02723638.2024.2329401; Federico Cugurullo, “Urban Artificial Intelligence: From Automation to Autonomy in the Smart City”, Frontiers in Sustainable Cities 2 (July 2020): 38, https://doi.org/10.3389/frsc.2020.00038; Federico Cugurullo et al., “The Rise of AI Urbanism in Post-Smart Cities: A Critical Commentary on Urban Artificial Intelligence”, Urban Studies 61, no. 6 (2023): 1168–82, https://journals.sagepub.com/doi/10.1177/00420980231203386.

This Mediapolis dossier adopts “the synthetic city” as a multivalent provocation for examining and interpreting the emerging implications of AI and related digital technologies for urban living – complementary to debates around “urban AI.”⁵This provocation was first put to participants at The Synthetic City Conference, co-hosted in September 2023 by the ECREA Media, Cities and Space Section and Dublin City University’s School of Communications. This dossier is inspired by that event, but it is not a collection of conference proceedings. While earlier versions of two dossier essays were presented at the 2023 conference, all contributions here were commissioned through a separate October 2025 call for papers. Rather than foregrounding technological determinism or dystopian critique, this dossier explores how agency, learning, temporality and banality are mutually reconfigured through situated practices of human-machine synthesis. In its most immediate sense, “synthetic” points toward AI technologies imitating, simulating or even replacing human agency in some areas of urban life. Generative AI outputs arguably shift how, for example, architects or urban planners imagine, visualise and describe urban spaces, introducing opaque computational agency into workflows and techniques of visual manipulation and production.⁶Raphaël Millière, “Deep Learning and Synthetic Media”, Synthese 200, no. 3 (2022): 231, https://doi.org/10.1007/s11229-022-03739-2; See also: Diogo P. Henriques and Michael Ostwald, “Generative AI, Architecture and Mathematics: Supporting Sustainable Development Goals”, in Handbook of Mathematics for Societal Challenges, ed. Bharath Sriraman (Springer, 2026), https://doi.org/10.1007/978-3-032-02376-6_20-1. Beyond generating urban representations, AI models also increasingly rely on synthetic data (e.g. artificially-generated images, records, driving environments) for their training.⁷Peter Lee, “Synthetic Data and the Future of AI”, Cornell Law Review 110, no. 1 (2025): 1–74. Matters get murkier still with the application of machine and deep learning systems to urban governance. As Orit Halpern and Robert Mitchell argue, the promise of emerging “smart” systems is to bring forth continuous, self-regulating processes of adaptation between data-generating and processing users, environments and machines, transforming urban management practices and even heralding the arrival of an “uncanny double of democratic decision-making.”⁸Orit Halpern and Robert Mitchell, The Smartness Mandate (The MIT Press, 2022), 71.

Yet the conventional invocation of “synthetic” – referring to simulation, imitation or unreal – is limiting. In a recent intervention, M. Beatrice Fazi suggests “synthesis” as a more appropriate connotation of synthetic.⁹M. Beatrice Fazi, “The Computational Search for Unity: Synthesis in Generative AI”, Journal of Continental Philosophy 5, no. 1 (2024): 31–56, https://doi.org/10.5840/jcp202411652. Pushing back against over-attention to the outputs of generative AI, Fazi stresses how internal pattern-driven and inferential logics in Large Language Models (LLMs) integrate, combine and unify distributed representations and information into distinctly new wholes. In a sympathetic, critical response, Shane Denson agrees with Fazi’s emphasis on synthesis, but stresses how synthesising AI models such as LLMs are always-already thrown into further synthesising dynamics with social and mediated contexts.¹⁰Shane Denson, “On the Very Idea of a (Synthetic) Conceptual Scheme”, Philosophy & Digitality 2, no. 1 (2025): 115–32, https://doi.org/10.18716/PD.V2I1.11664. On this point, Denson evocatively suggests ““we could say that although it is we who “prompt” LLMs to produce new texts, those outputs in turn serve as “prompts” for us to provide them with sense. All the meaning-making remains squarely on the side of the user/reader” (p. 116); See also Louise Amoore et al., “Politics of the Prompt: Government in the Age of Generative AI”, Economy and Society 54, no. 3 (2025): 573–96, https://doi.org/10.1080/03085147.2025.2560177, who discuss how practices of prompting, as acts seeking to elicit certain model behaviour, create novel conditions for political action and governance. The synthetic city, on this score, is not so much an inhuman, unworldly, or separated-off machine-generated urbanity. Rather, it suggests we understand AI models as novel cognitive infrastructures of urban life, management and imagination, with human users as fully-fledged participants.¹¹Benjamin Bratton and Blaise Agüera y Arcas, “The Model Is the Message”, Noema, 12 July 2022, https://www.noemamag.com/the-model-is-the-message/. It follows that “synthetic” and “synthesis” are not competing alternatives: when one looks closely, the former always entails the latter.

To be sure, the essays in this dossier deploy “synthetic” and “synthesis” in their own diverse valences. But what they hold in common is resisting the temptation to try and catch up with, or even somehow get ahead of, all the myriad developments associated with “AI.” Rather than speed up, they slow down, seeing AI not in isolation, but as interwoven with longstanding and geographically differentiated interdependencies of digital technologies and urban life. This highlights many concerns (including, for some, questions around political economy, commodification and labour). But in what follows, we portray this dossier as most prominently asking questions about agency, learning, temporality and, in the end, the banality (though not triviality) of synthetic cities.

Where is agency in synthetic cities?

It may be unsurprising to find this dossier engaging first and foremost with questions of agency in the synthetic city. In his contributing essay, Leandry Jieutsa traces a drift from digital governance to “self-operating governance.” Drawing on scenario-based interviews with eleven city managers in the Canadian province of Quebec, he describes how ordinary bureaucrats find themselves intervening not so much at the fine grain of routine decision making, but increasingly supervising semi-autonomous algorithmic systems and infrastructures. He concludes that, while not yet in place, fully autonomous urban governance is emerging through optimisation pressures, institutional constraints and socio-technological trade-offs.

Luke Hespanhol, Michiel de Lange and Nanna Verhoeff argue that the most simplistic implementations of such urban AI systems amount to a civic crisis. They describe a “triple actionability gap” where: autonomous decision-making transpires without meaningful human involvement; urban citizens become extractable resources rather than users, surreptitiously rendered into the data points that train AI systems; and the development and implementation of such systems are too often expert-led, and prone to excluding ordinary civic voices and public values. Hespanhol et al. insist that civic agency must be learned with AI rather than simply delegated to it, proposing “Civic Urban AI” as a lens for collective learning, reframing the debate from ethics-as-compliance to an ongoing and situated political pedagogy.

There is an evident tension between these two accounts. When presented with a hypothetical scenario relating to an autonomous fire management system, Jieutsa’s Quebec-based city managers unequivocally rejected complete AI system autonomy, stressing the need for “humans in the loop.” Yet for Hespanhol et al., human actors – from decision-makers to ordinary citizens – nevertheless appear to be pushed out by stealth, hence the emerging civic crisis. Navigating the apparent discrepancy between these two accounts might be less a matter of deciding which one is right, and more about understanding differences in how human and computational agency gets attributed in practice.

In their account of Tonomus, the subsidiary company charged with building the “cognitive-city” infrastructures for the Saudi Arabian NEOM megaproject, Mehdi Ghassemi and David Murakami Wood show how such attributions of agency work in situated practice. Their discussion of M3LD, NEOM’s intended consent-management platform, is particularly revealing in this regard. M3LD promises a system for users to monitor the use and ownership of their data, and control their profile and privacy. While the public launch documentation deploys humanistic terms such as “control” and “profile,” Ghassemi and Wood describe how this ascribes agency based not on broad rights for NEOM “citizens,” but more narrowly, by pre-determining users as legal and technical pivots within M3LD data flows.

Miguel Valdez offers another, distinct story about agency. His essay seeks to counter assumptions that AI principally entails the arrival of superhuman urban intelligence, discussing embodied entanglements with actually-existing AI robotics such as grocery delivery devices. Accounts emphasising highly intelligent AI technologies tend, Valdez argues, to take agency away from humans to act and even to create (or co-create) meaning; in other words, humans become largely passive users of autonomously-managed urban environments. Yet when we look at already deployed, task-orientated robots of more limited intelligence, accounts of humans and computational technologies sharing space, emotional resonances, collaborations – but also safely ignoring one another – become more possible. He argues that such “silent dialogues” between humans and computers constitute emerging, lived experiences that “give rise to new meanings and knowledges.” Valdez stresses that the presence of narrowly intelligent robots does not negate the possibility of a city brain: networks between deployed robots and distributed intelligences can be fluid; and putting them into a dichotomy is a lingering Cartesian dualism that “can be enforced by policy or by design, but … is not a natural law of AI.”

What these contributions show, then, is that the most helpful question may be less “how much agency do urban AI systems have?” and more “how does agency get attributed in and around urban AI?” – whether by software developers, policy- and decision-makers, urban inhabitants or us as scholars. Asking that question might demand we refuse a human/machine binary altogether. One notable and well-conceptualised example of such a refusal is N. Katherine Hayles’s “integrated cognitive framework,” which she has most recently deployed for thinking about AI.¹²N. Katherine Hayles, Bacteria to AI: Human Futures with Our Nonhuman Symbionts (University of Chicago Press, 2025), 2. This framework extends cognition beyond humans to include all forms of biological life, as well as computational media such as LLMs. Once it is no longer reserved for humans alone, “cognition” becomes disentangled from Enlightenment ideas and ideals of the human mind and becomes visible as “cognitive assemblages” inherently produced through the interaction of human and non-human actors.¹³It is worth adding here that other strands of debate around posthumanism underscore how the same Enlightenment ideas and ideals rest on unstated assumptions of “the human” as white, European, male, heterosexual, able-bodied, and propertied; see Rosi Braidotti, The Posthuman (Polity Press, 2013). A similar idea has been proposed by one of our contributing essay authors, Luke Hespanhol, with his notion of “human-computer intra-action,”¹⁴Luke Hespanhol, “Human-Computer Intra-Action: A Relational Approach to Digital Media and Technologies”, Frontiers in Computer Science 5 (August 2023): 1083800, https://doi.org/10.3389/fcomp.2023.1083800. which argues humans, computers and worlds are ontologically inseparable. The upshot is that measuring the agency of urban AI systems may be impossible. The more viable and useful task, broadly undertaken across the contributions to this dossier, is to study, conceptualise and self-reflect on how agency is discursively and technically attributed in AI-mediated urban life.

Synthetic city learning and localisation

One of the most tractable themes around which this dossier’s contributors pose questions about agency is learning in synthetic cities. Colin McFarlane’s seminal book Learning the City¹⁵Colin McFarlane, Learning the City: Knowledge and Translocal Assemblage (Wiley-Blackwell, 2011). provides a useful reading of urban learning for our purposes: as the processes whereby distributed knowledge gets translated and reconstructed by intermediaries that are always themselves situated. On his account, urban learning entails localising translocal assemblages of knowledge, mediated by human and non-human actors. Scholars such as Halpern have adopted a broadly similar view in thinking about how AI and more generally smart systems, which entail distinctive combinations of human and machine learning, make for novel tensions between contingent and automated decision-making processes.¹⁶Orit Halpern, “Epilogue: Beyond Data and Crisis”, in Data Power in Action: Urban Data Politics in Times of Crisis, ed. Ola Söderström and Ayona Datta (Bristol University Press, 2024).

This chimes with Jess Reia’s contribution around the distinctive questions the urban night poses for emergent AI-mediated urban learning. Reia points out that night-time urban governance relies on an abundance of data captured from humans, non-humans and sensors that can be efficiently processed and interpreted through “surveillance processes [that] do not capture and understand the complexity and nuances of the night.” They argue for a broader humanistic learning about the night, which must include messier but no less valid or meaningful data, such as music scenes, food consumption, sex work or undocumented labour. Such data is defined by “partial visibility, refusal, and ambiguity,” which according to Reia is better captured through bottom-up approaches such as oral history, interviews and focus groups. Their point is not so much to oppose “human” and “machine” learning about the urban night, but rather to remind us that data is not found: it is always made, demanding care in the adoption of AI systems for knowing as well as governing urban life.

Such care is clearly offered by Eric Gordon, Loretta Lees, Carolyn White and Chang Xiao in their account of conceptualising and designing a potential AI agent-enabled urban digital twin (UDT) of the Fenway neighbourhood of Boston. Their intended Fenway UDT is distinctive, they suggest, in how it would link AI agents with 25 corresponding neighbourhood locations, with each agent relying on place-specific knowledge derived from structured community conversations convened by the research team, and a range of heterogeneous extant cultural data, challenging conventional, quantified and efficiency-driven notions of UDTs. Built using an approach called “agentic retrieval augmented generation” (agentic RAG), these AI agents would retain place-based data as a separate and retrievable archive, rather than using it as a corpus to train a model, making it easier to update and its provenance more visible. Their speculative design is an interesting test case in AI-enabled urban epistemology, where what it means to know a city and who gets to define that knowledge is kept firmly in view.

This reminds us, however, that the mere physical geographical localisation of AI models and systems might be insufficient for navigating tensions between human and machine learning. Hespanhol et al. point out that it is perfectly possible for a large-scale AI model to be trained to learn about locally-specific conditions, but for that learning to then primarily profit the translocal company owning the model. In their essay, they point to a counter-example of indigenised AI learning, where researchers created a speech recognition model capable of transcribing interviews with rapidly declining te reo (Māori language) speakers, a project not only useful for, but owned by, the local communities it serves.

Joel McKim’s essay on the varied adoption of generative AI in the architectural field, on the other hand, highlights the peculiar learning embodied in AI image models. Trained on diverse images from art, photography, visualisations, advertising and entertainment, these models mathematically abstract such visual material across latent spaces capable of generating new images based on language prompts. It is an abstraction that collapses “periods and mediums, high and low culture, realism and fantasy,” replacing such distinctions with statistically probable visual similarities. Quoting media theorist Roland Meyer, McKim points out how this flattens out historical context, affording architectural practices a “generic pastness.”¹⁷Roland Meyer, “Generic Pastness: AI Image Synthesis and the Virtualization of the Archive,” research seminar, Bibliotheca Hertziana – Max Planck Institute for Art History, Rome, October 13, 2023, https://www.biblhertz.it/events/35511/2719701 (accessed September 6, 2026). While this might open the door to more conservative architectural visions or designs – and at the extreme, reactionary projects such as Praxis – McKim points out how some practitioners have used the same technologies to visualise cities and urban life in new and critically challenging ways.

In their discussions of phenomena such as latent space or agentic RAG, McKim and Gordon et al. emphasise the specific critical potential of understanding AI models via their technical specificities. Hespanhol et al. agree that “explainable AI” might be valuable, but also suggest such a move has limits when it comes to addressing the civic crisis that concerns them. They argue it might be unreasonable to expect average urban citizens to become literate in such complex systems, in the same way that we would not expect explainable electricity as a precondition for using such mundane infrastructures. As we will see later, it is in any event difficult to be literate about infrastructures that appear to be banal or recede entirely from perception. Irrespective of these differences, this dossier’s discussions around synthetic city learning show, as McKim expresses well, that like history in general, what AI models learn has a determinative bearing on what urban futures are thinkable and speakable.

The temporal politics of synthetic cities

If there is a distinctive politics to the synthetic city, the contributions to this dossier appear to suggest that it is temporal. In varied ways, each of the papers points to how AI models crystallise in advance, or anticipate, what is later knowable or actionable in urban life. Jieutsa, for example, describes “self-operating governance” as a kind of feedback loop in which AI-optimised operational infrastructures help train predictive models, and predictive models then further influence automated operations. Human actors are not completely excluded from such anticipatory processes but repositioned as supervisors rather than producers. Reia stresses how the urban night in many ways resists the anticipatory techniques of AI and machine learning, which lack an understanding of “messy” and complex urban contexts. But we can notice here the temporal conditions of such resistance, where the practices in communities making urban nightlife distinctive and valuable are threatened by the predeterminations of technologies such as predictive policing.

Fran Meissner’s account of the nexus between cities and orbital infrastructures makes this temporal politics exceptionally clear. On first glance, Meissner’s account might seem to highlight a spatial rather than temporal politics: synthetic cities on the Earth’s surface being inextricably tied to far-flung Earth-observing (EO) satellites in orbit. But the significance of EO infrastructures is in how they can sense and detect Earthly phenomena, and how this helps determine what can or cannot subsequently be matters for urban intervention. These geographically-uneven capabilities provided by ESA, NASA or more recently private companies are moreover increasingly prefigured by “Edge AI” technologies built directly into orbiting satellite objects, allowing for ever-more-timely data processing and transmission.

We can also return here to urban digital twins and the broader twinning paradigm, since they promise new kinds of capacities for predicting urban futures, combining generative AI with sensor data, actuating infrastructures and institutional bureaucracies that datafy, model and act in urban environments.¹⁸Christoph Borbach et al., “Making Everything Ac-Count-Able: The Digital Twinning Paradigm”, New Media & Society 27, no. 8 (2025): 4369–84: https://doi.org/10.1177/14614448251338289. Two essays in this dossier centre on digital twins that effectively do not and may not ever exist. Ghassemi and Wood write about a digital twin envisioned for NEOM, an urban megaproject for which the contracts are being cancelled and scaled back at the time of writing. Gordon et al. describe their own design and conceptualisation of the Fenway UDT, expressed at once as something that could be built, yet at least for now, a work of speculative fiction. But what is significant about these essays describing unbuilt digital twins is that they both concern situations where technical parameters establish and authorise in advance subsequent action. Yet they do so from notably different positions, describing very different power dynamics. Ghassemi and Wood point to a corporate megaproject where technical atmosphere, consent and tenancy are pre-specified before any residents even arrive; Gordon et al. describe a participatory process where community deliberation would give shape to 25 AI-enabled sites, each incomplete by design.

The epigraph to Gordon et al.’s essay, from Jorge Luis Borges’s “On Exactitude in Science,”¹⁹Jorge Luis Borges, “On Exactitude in Science,” in Collected Fictions, trans. Andrew Hurley (New York: Penguin, 1998), 325. reminds us of the hubris that can characterise efforts to completely map and mirror geographical worlds such as cities. As they point out, a hypothetical cartography that rendered a city so exactly that there would no longer be any distinction between the place and the map would lock space to time in such a way that nothing could ever change. While their epigraph warns that efforts toward completeness entail foreclosure, their speculative design argues that neither completeness nor foreclosure is inevitable. While AI-enabled systems such as urban digital twins risk the establishment and imposition of highly uneven and unequal means for capturing urban life, they also present possibilities for building more dynamic kinds of urban representations around a multiplicity of urban futures. McKim makes a similar point in his discussion of imaging models in the architectural field: yes, they pose the risk of politically and financially conservative architecture; but also make possible new visions of the city. However difficult it may be to accomplish, there remains scope to reimagine what AI models and systems learn from, and thereby open new possibilities around the kind of urban futures that can be claimed.

Conclusion

What the contributions to this dossier do well is slow down, seeing AI not as a sudden break, but an instance in the longstanding interdependencies of digital technologies and urban life. In so doing, this dossier arguably emphasises the banal, rather than spectacular synthetic city. But banality should not be mistaken for triviality. In his seminal book Banal Nationalism,²⁰Michael Billig, Banal Nationalism (Sage, 1995). Michael Billig described how established, wealthy Western nations maintain national identity through everyday, routine and often overlooked habits, language and symbols. One of his best examples is “the unwaved flag”: flags hung on public buildings, shown on stamps and currencies, stitched onto shirts or backpacks. Banal phenomena such as these are not benign leftovers of nationalism, but rather the subtle means through which it is made mostly normal and unexceptional, yet also available for overt mobilisation. Ghassemi and Wood show something along these lines with NEOM: an overt promotional video and a quiet corporate sector page are two forms of the same claim, and the second is why the first can return.

A parallel in the field of computing is Mark Weiser’s influential observation, frequently cited in media and technology studies,²¹For example, see: Paul Dourish and Genevieve Bell, Divining a Digital Future: Mess and Mythology in Ubiquitous Computing (MIT Press, 2011); Mercedes Bunz and Graham Meikle, The Internet of Things, Digital Media and Society (Polity, 2017). that “The most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it.”²²Mark Weiser, “The Computer for the 21st Century”, Scientific American 265, no. 3 (1991): 94. Weiser was putting forward a kind of philosophical point that computation would become profound once it disappeared, or ceased to be so conspicuous. Yet this was less prophecy than design goal, realised via a corporate research programme at Xerox PARC. Receding into the banal, in other words, can be engineered. So, the banality of the synthetic city is far from politically neutral. Its everydayness facilitates capital accumulation at scale, making questions of ownership, profit and dispossession opaque.

The contributions to this dossier seem to suggest that if the synthetic city has already arrived, it has not only been via occasionally spectacular announcements or megaprojects, but a multiplicity of smaller practices and technologies, spread across technologically-mediated urban life and its governance. Ghassemi and Wood’s paper describes how NEOM’s high-profile Tonomus brand has receded, replaced with a more mundane and technical corporate sector page that “still carries the cognitive-city premise, but at lower volume.” Meissner’s far-flung EO satellites are not sunk into the synthetic city, but still enable it through their silent hovering. Jieutsa provides an account in which onetime AI innovations become unremarkable environments of municipal bureaucracy. Valdez draws our attention to the narrowly-intelligent “icebox on wheels” as one defining form of actually-existing urban AI robotics, rather than an omnipresent city computer, reminiscent of HAL 9000 from Stanley Kubrick’s 2001: A Space Odyssey.

A similar endeavour is evident in the recent AI in the Street project,²³Dominique Barron et al., AI in the Street: Lessons from Everyday Encounters with AI Innovation (Careful Industries, 2025). led by Noortje Marres, which has tried to grapple with how people make sense of the AI-based technologies encountered in everyday urban environments. While close scholarly analyses might find AI technologies clearly present or acting in the streets of many cities, the nature of that presence is often indecipherable and even invisible to everyday urban dwellers. AI in the Street entailed design-based interventions in the form of everyday AI observatories in four UK and one Australian city streets. In the process of running these observatories, one of the challenges the project team encountered was just how obstinately inaccessible AI infrastructures can be for ordinary street dwellers and users.²⁴Alex Taylor et al., “Reciprocity Deficits: Observing AI in the Street with Everyday Publics”, CoDesign, 12 July 2026, 1–18, https://doi.org/10.1080/15710882.2026.2697269.

This reminds us that there are perhaps two withdrawals at play in the synthetic city. First, AI systems withdraw and become banal, operating below the threshold of everyday notice, receding into phenomena like corporate sector page vocabulary or sensors embedded in streetside furniture. Second, urban dwellers also attempt to withdraw from capture by these systems. Yet the first withdrawal is what makes the second so difficult. As we mentioned in our earlier discussion of urban AI learning: it is very difficult to have an actionable and critical technical literacy of AI systems that are banal, or which recede from daily perception. Ghassemi and Wood suggest that withdrawal from urban AI systems is less a matter of logging off than addressing how we inhabit cities already infused by AI systems. This is a difficult task to navigate, but one for which we hope this dossier begins to provide some means of wayfinding.

Scott Rodgers

Scott Rodgers is Audio Editor for Mediapolis, where he hosts the Mediapolis Now podcast. He is Reader in Media and Geography in the School of Creative Arts, Culture and Communication at Birkbeck, University of London, and co-director of Birkbeck’s Centre for Creative AI. His research works at the interface of media and geography, in recent years on the urban and local geographies of social media and digital platforms, and increasingly on how AI is becoming the next mediating infrastructure of everyday urban life. This concern with the ordinary character of mediation runs across his work, including Conditions of Mediation: Phenomenological Perspectives on Media (Peter Lang, 2017), co-edited with Tim Markham.

Helena Atteneder

Helena Atteneder is a senior lecturer at the Institute of Media Studies at the University of Tübingen in Germany. Her research is interdisciplinary, combining media and communication studies with human geography. Specifically, she examines the relationships between media, places/places, mobility, and power, with a particular emphasis on methodological advances. Her recent work includes ‘Maps as Media Constructs’, ‘Digital Placemaking in Salzburg (Austria)’, ‘Geospatial Ideologies of ESRI Blog Posts’, and ‘The Diverse Media and Mobility Practices of Commuters’.

You do not have any posts.

Lou Therese Brandner

Lou Therese Brandner is a postdoctoral researcher and team leader at the International Center for Ethics in the Sciences and Humanities (IZEW) at the University of Tübingen. Her research revolves around ethical and societal aspects of technological developments (data-based technologies, AI, robotics) in a diverse range of contexts, such as surveillance, education, (digital) labour and sustainability. She is Vice-Chair of the ECREA Media, Cities and Space section.

You do not have any posts.

Marcos Dias

Marcos Dias is an Assistant Professor in the School of Communications, Dublin City University. His research expertise focuses on the social and spatial impact of communication technologies and urban infrastructure on citizen well-being. He is the author of The Machinic City: Media, Performance and Participation (Manchester University Press, 2021).

You do not have any posts.

Diogo P. Henriques

Diogo P. Henriques is an independent consultant, creative technologist, designer, and researcher based in Barcelona, Spain, working for international organisations, universities, and practices in Europe and Asia. Recently, he has been with the Open Data Charter and the Inclusive Policy Lab of UNESCO, working on projects funded by Erasmus+ and Creative Europe Programmes of the European Commission, among others. Since 2024, he has been the Communications Officer for the Media Cities and Space section of ECREA, and, since 2019, corresponding editor of the Nexus Network Journal and of the book series Mathematics and the Built Environment, published by Springer.

You do not have any posts.

Sara Trentham-Black

Sara Trentham-Black is a PhD student and lecturer at Sheffield Hallam University, UK. Her research interests include cultural governance, datafication, technology, and place. Her PhD explores the impact of datafication on small cultural organisations and how organisations navigate the relationship between funding, policy and data. She has also worked with City of Doncaster Council as a critical friend to the delivery of their cultural strategy, and is part of a multidisciplinary team delivering a three-year project to develop learning about Creative Health Boards.

You do not have any posts.