Research paper: artists working with GenAI’s limitations

XTREME researcher Justyna Starostka presented the paper From Malfunction to Method: Artistic Strategies for Working with GenAI’s Limitations, co-authored with Anders Sundnes Løvlie, Leo Vitasovic, Willem van der Maden and Sami Sebastian Brandt from the IT University of Copenhagen, and Laurence Cliffe and Paul Tennent from the University of Nottingham.

Based on interviews with 13 contemporary artists, the paper explores how artists use the limitations and malfunctions of generative AI as creative material, rather than treating them as flaws to be fixed.

Read the full paper (open access): https://doi.org/10.1145/3829807.3829915

The Dissonance Workshop: performing with the glitch

At the conference, Justyna also took part in The Dissonance Workshop: Live Performance and Generative Friction with AI. Combining analog activities, discussions and digital explorations, participants embraced the glitches, frictions and malfunctions of GenAI and turned them into artistic material.

The result was the audio-visual performance Beavers in the Casino, presented as part of the conference opening reception. In a sense, the performance put the paper’s findings into practice: treating malfunction not as an obstacle, but as the work itself.

You can read more about the workshop and watch a screen recording of the performance at https://dissonance.craitr.no

Many thanks to fellow workshop participants Danny Snow, Enrique Encinas, Eugénie Desmedt, Nadine Jachmann and Daniëlle Leentjens.

Photo: Wendy Zhou

 

Related articles

XTREME research release: ExMR motion capture model now open to everyone

ExMR, a new open-source model developed within XTREME, captures 3D body, hand and foot motion from a single video. The code and a browser demo are now available, and creators are already building on it.

Stepping inside Vivaldi’s Four Seasons with the XTREME Living Lab

This autumn, the XTREME Living Lab invited audiences in Limerick to step inside Vivaldi’s Four Seasons in mixed reality, at Culture Night 2026 and on the University of Limerick campus.

XTREME research at ECCV 2026, including a Best Paper Award

XTREME-funded research featured in five papers at ECCV 2026 in Malmö, one of the world’s leading AI and computer vision conferences. Among them, OvDSGG by John Lu Helsby and Yi Yang won the Best Paper Award at the Contextus workshop.