Information About Our Guests

Manuela Naveau, PhD is a university professor, independent curator, and art-based researcher. Over the course of nearly 18 years at Ars Electronica, she served as Head of Ars Electronica Export, co-developing the department and overseeing its operations from its founding onward. Since 2020, she has been Professor of Interface Cultures / Critical Data at the University of Arts Linz, Austria. In February 2023, she assumed leadership of the department Interface Cultures and founded the Critical Data Research Group.

UNBEHAGEN - When Critical Data Touches the Shadow of the Invisible

In her keynote address “UNBEHAGEN — When Art Uses Data to Touch the Shadow of the Invisible,” Manuela Naveau explores how artistic research can go beyond merely visualizing data to contribute to a critical engagement with the systems, imaginaries, and power structures that shape our world. Her perspective opens up new ways of engaging with data—not merely as a technological resource, but as a cultural, political, and aesthetic condition of our present.

Joanna Zylinska is a writer, artist, curator and Professor of Media Philosophy + Critical Digital Practice at King’s College London, where she directs the Centre for the Ecologies of Attention and Perception. An advocate of ‘radical open-access’, she is a co-Director of Open Humanities Press and an editor of its MEDIA : ART : WRITE : NOW book series. Zylinska is an author of a number of books – including The Perception Machine: Our Photographic Future Between the Eye and AI (MIT Press, 2023), AI Art: Machine Visions and Warped Dreams (Open Humanities Press, 2020) and The End of Man: A Feminist Counterapocalypse (University of Minnesota Press, 2018). Her own art practice involves experimenting with different kinds of image-based media. Zylinska is currently researching perception and cognition as boundary zones between human and machine intelligence, while examining various narratives of collapse, be it on the level of AI models or planetary systems.

From Model Collapse to System Failure: Remaking Data for Better Worlds

Collapse has become one of the defining concepts through which the present is being narrated today. We encounter it in relation to climate systems, political institutions, economic infrastructures, AI models, social worlds and, increasingly, epistemic authority itself. Yet the question I want to pose in this talk is not simply whether collapse is really happening or how far advanced it might be. It is rather what kind of concept collapse has become: what kinds of affects it mobilises, what kinds of subjects it presupposes and what kinds of worlds it still allows us to imagine. Starting from the recent technical discourse on model collapse in generative AI, where recursively trained systems begin to lose low-probability variation within its data and produce increasingly generic outputs, I will ask how this concept might be rethought within the arts, humanities and social sciences. Model collapse will be positioned in this argument as a diagram of cultural and political narrowing, a condition in which systems feed on what they have already processed, optimised and made probable. Yet it will also be seen as an opening within the all-too-rigid models of understanding human intelligence and subjectivity. Moving between aesthetics and ethics, the talk will consider what it might mean to move from diagnosing model collapse within machine-learning systems to recognising broader forms of system failure within our real-world ecologies. We will also interrogate how data – and data worlds, with all their material infrastructures – might be remade otherwise, through practices of critical attention and creative recomposition. The wager is that collapse, if thought carefully, may open a space for imagining better worlds beyond the exhausted logics of averaging and optimisation.