GENERATED/-IVE
FUTURES

This presentation examines the risks generative AI poses to democratic societies, focusing on its role in amplifying challenges related to disinformation and other forms of misleading content within the information landscape. A central argument is that generative AI fundamentally alters the conditions for assessing the trustworthiness of information by making sources increasingly invisible. As AI systems reduce the visibility of information provenance, the presentation asks whether we are moving from a state of crisis to a post-crisis of information.

As engagement with traditional news media declines, social media platforms increasingly shape how news is produced, circulated, and consumed. In this environment, the boundary between "true" and "fake" information is becoming less distinct, while emotionally charged and AI-generated content can spread more rapidly than verified news reporting. This presentation examines the challenges posed by AI-generated multimodal news content, including deepfakes and manipulated videos. It also presents ongoing research that uses AI methods to analyse problematic news narratives in large-scale datasets of short-form news videos, with a particular focus on TikTok.

This talk focuses on the shift in platform governance from moderating individual pieces of illegal or harmful content toward managing systemic risks on social media platforms, in an era of generative technologies and algorithmic amplification. The Digital Services Act is the newest and most ambitious regulatory framework targeting this development, and in the final stages of the legislative process, a crisis response mechanism was added, granting the European Commission the power to require changes to platform operations during extraordinary societal crises. While the implications of this mechanism remain uncertain, it raises important questions about power, accountability, and institutional balance in shaping online information environments in a present – and future – increasingly characterized by recurring risks and crises.

AI practitioners and stakeholders are rather keen on putting forward predictions for the future of AI and of our putatively AI-infused future societies. These exercises in speculative prediction seem more common (or, at least, more prominent) in AI than in other fields of science and technology. In this talk, I will explore why this is so. I suggest that AI has historically invited an unusual blend of sci-fi inspired hopes and terrors, marketing interests, and more or less conscious reflection on the limits and failures of humans. I will also examine some of the epistemic challenges involved in such predictions.

Technofascism does not predict the future; it builds one. I argue that reactionary modernism, the marriage of romantic unreason and technological power, has fused with the speculative fictions of Silicon Valley into a perverse worldmaking project: a civilisational horizon in which only the technofascist oligarch harnesses the power to speculate, and his 'supremacy' is made to feel not merely desirable but inevitable and righteous. In the terms of Critical Fantasy Studies and ontological security, this fantasy binds us through enjoyment rather than belief. Algorithmic capitalism streamlines and normalises that enjoyment across its social media platforms and attention-harvesting technologies, until reactionism is experienced and felt as 'common sense'. That normalisation is neoliberalism's own work, revealing the far-right not as antagonistic to it but rather its mutation. Their co-constitution in technofascism is dangerously and actively foreclosing - both in our imaginaries and material realities - all futures but the white supremacist one.

How do Google Images and visual generative AI make urban futures visible, thinkable, and desirable? Drawing on a visual study of Google Images and AI-generated images, this talk examines recurring tropes of green techno-urbanism, smart infrastructures, seamless mobility, and depopulated sustainability. I argue that these images do not simply represent possible futures; they actively constrain the aesthetics of the possible by privileging commercially viable, techno-solutionist imaginaries while marginalising social complexity, participation, postcolonial perspectives, and everyday urban life. The talk asks how critical visual literacy might open space for more plural, democratic, and situated urban futures.

This contribution will discuss the political economy of AI capitalism, arguing that treating AI as a General-Purpose Technology (GPT) reveals how Big Tech's dominance is built on data commodification, extraction, and concentrated control over computing infrastructure and AI talent. This winner-takes-all dynamic drives monopolisation and digital enclosure, concentrating power among a handful of corporations. Drawing on critical political economy, the paper proposes the commons as an alternative framework — encompassing common data, shared compute capacity, and public investment in AI expertise — to democratise AI development and distribute its benefits more equitably across society.

Regulatory dynamics underlying the EU's digital transformation limit the effective exercise of civil society's counterpower. Recent initiatives such as the DSA and AI Act promote civil society representation in multi-stakeholder governance, yet structural inequalities in resources and recognition obstruct meaningful participation. Meanwhile, EU policies promote the digitisation of everyday infrastructures, where people's data gain economic, political, and representational value for industry and the state. These processes privilege the quantifiable, visible, or legible over other kinds of knowledge, so the EU's politics of visibility risks undermining alternative forms of resistance and participation. Drawing on Édouard Glissant's right to opacity and Clare Birchall's radical secrecy, the talk asks how communal approaches to data infrastructure governance might give civil society greater agency over the societal use of AI.

While things have always had a kind of agency and participated in relations, the artificial agencies now at play in contemporary data-intensive environments are unprecedented. Using everyday digital things that are connected to data-driven platform ecosystems entails participating in multiple platformed relations that are typically optimized to generate value for other platform actors. Although exploitative configurations of platformed relations may be common, they are not inevitable. This talk makes a case for regenerating platformed relations – recognizing them as ultimately produced by humans and thus possible to reconfigure. Alternative formations can be designed that are tuned in relation to human agencies, both individual and collective, and the larger ecological entanglements of diversely livable futures.

Memory has become a unique selling point for AI companies. According to the creators of Dume.AI, a popular 'AI executive assistant', "memory has become the critical feature that separates a tool from a true assistant." Positioning their products as fundamentally different from earlier chatbots, creators of what I call 'mnemonic AI' claim that such systems better understand who their users are, what they deem valuable and what their routines are, based on retrieval-augmented generation (RAG). I will demonstrate that claiming that an AI model 'has memories of you' is a rhetorical and political move. By attributing and integrating an essential, 'useful' and seemingly neutral cognitive capacity to their systems, AI companies are well on their way in creating new forms of dependencies and technological lock-in.

You and I, as users of digital media, have a digital afterlife, whether we want it or not. In recent years, developments in AI have brought new possibilities, but also challenges, for studying our digital afterlives as AI dead: biologically dead, virtually alive, and socially active beings (Bassett 2022), and for examining the posthuman relationships between the dead and the living. This presentation sheds light on the ontological and methodological experiments in developing research on the AI dead and reflects the ethical and political implications this rapidly evolving study field poses for a society characterized as post-mortal.

The emergence of new forms of artificial intelligence, such as large language models (LLMs), has a profound impact on collective remembrance. LLM-based applications increasingly serve as sources of historical information for the general public worldwide; however, the implications of their probabilistic logic — together with the potential alignment with values associated with national memory regimes — for the articulation or erasure of specific aspects of the past remain unclear. In this talk, I will discuss how we can study such alignment in the context of historical information and its implications for projecting the future, using Holocaust remembrance as a case study.

Generative AI is changing online platforms in ways that challenge platform governance and enforcement. Systems that can generate content at scale create new risks related to criminal content, civic discourse, and public health. This presentation explores how these developments relate to the Digital Services Act (DSA), with a particular focus on systemic risk, platform responsibility, and what opportunities the DSA offers to researchers.

As the use of Gen-AI increasingly permeates our societies and educational systems, its ubiquitous use to summarise, draft and rephrase poses a number of crucial questions for the integrity of various political and legal processes. This is particularly salient in the field of public international law, the body of rules that broadly regulates how members of the international community interact with one another. Faced with major challenges such as climate change and renewed inter-state conflicts, international law simultaneously appears as particularly vulnerable to a quiet takeover by Gen-AI due to its structure, the nature of its rules and the asymmetry in means between its subjects.

This presentation conceptualises AI chatbots as a site of slow violence: harm that manifests in gradual, attritional, and often imperceptible ways. The relational modality of chatbot engagement enables harms that unfold over time, through sustained interaction, emotional proximity, and adaptive sycophantic responses, shaping users' epistemic environments, political agency, and social trust. These attritional harms also expose structural limits in current responses: instruments such as the EU Digital Services Act and the AI Act acknowledge systemic risks, yet a human rights framework oriented toward identifiable, episodic violations struggles to capture the cumulative erosion of dignity, autonomy, and the preconditions for democratic deliberation.

Generative AI is often described through a misleading opposition: either as a tool we use or as a mind we confront. Both vocabularies obscure the most socially significant dimension of these systems, namely that we engage with them relationally and are mutually shaped in the encounter. Drawing on my recent work applying symbolic interactionism to generative AI, I offer a third vocabulary for understanding it as a relational other. Through the concepts of conversational choreography and the socio-technical self, I show how meaning and a transient, processual sense of agency emerge through iterative exchange, and how futures are not retrieved from these systems but rehearsed and shaped with them.

AI bureaucracy promises more efficiency, better service and more just decisions. But what happens when artificial intelligence is moving into the welfare state's machinery? When people are meeting the digital platform of the Insurance Agency, chatbot Dolly of the Employment Services or Trelleborg municipality's decision robot? Is AI really the better bureaucrat? Or do we just exchange human constraints with blindness when we leave Max Weber's iron cage for the algorithmic black box?

Workplace datafication is increasingly enacted through AI-driven systems that monitor productivity, automate performance evaluation, and profile employees in real time. Drawing on qualitative fieldwork from the Datafied Living project, with longitudinal interviews across knowledge work, the public sector, and service industries in Denmark, this paper examines how workers make sense of and negotiate organisational tracking. I use the concept of aspirational control to capture a recurring tension: while tracking is experienced as surveillant and at times exploitative, employees also invest normative expectations in datafication as a technology of 'good work', transparency, and optimisation. Drawing on welfare-state values of equality, trust, and co-determination, Danish employees produce forms of resistance and accommodation distinct from those in liberal market economies.

This introductory talk provides a brief overview of previous research and the developments that led to the formation of the Q.F.AI network. Earlier studies have highlighted several challenges in AI from gender and queer perspectives, including issues related to identity, privacy, surveillance, bias, and fairness. Against this backdrop, a series of events and collective actions since 2022 brought together an interdisciplinary group of researchers to develop a shared response. This led to the creation of Q.F.AI as a collective effort to address the risks and challenges associated with the increasing implementation of AI in society, while also exploring alternative queer futures of AI.

Algorithmic culture has the ability to give shape and form to previously unacknowledged or underexplored aspects of desire, but also tends to do so in the only way known by the machine, i.e. through stable and easily detectable categories. As such, it presents the user to ready-made highways expected to lead to her objects of desire, while at the same time often throwing her most unarticulated and messy inner terrains back at her. This calls for a rearticulation of queer phenomenology in times of everyday automation.

AI systems such as facial analysis and motion capture rely on labeled datasets to classify bodies into rigid categories, frequently reproducing cisnormative, racial capitalist, and ableist assumptions. The use of bodily data in automated systems has intensified forms of undesirable identification, such as AI-driven surveillance, reducing bodies into identity markers to be (mis)detected. Drawing on an exploratory study of a motion capture system, I challenge top-down approaches that rely on fixed, binary, and normative identity categories. From a critical and queer theory perspective, I explore the affective role of bodies in glitching and resisting AI systems' classificatory regimes.