Mobilizing Models: How Artistic Intelligence Reprograms AI

This panel examines the concept of the model at the intersection of art and AI, asking how artists engage competing notions of intelligence to challenge existing models and mobilize new methods. While artistic practices define, critique, and reimagine cultural, stylistic, and conceptual models, AI algorithms train, deploy, and optimize technical models to calibrate, predict, and generate outputs like tokens and pixels. Through this tension, the panel investigates how creative practices reconfigure dominant systems while imagining new possibilities.

Holly Herndon reprograms AI by training a model on community choirs' collective singing, foregrounding ritual and resistance to commercial uniformity. Mimi Onuoha critiques machine learning's narrow focus on reasoning and prediction, which excludes bodily and emotional forms of intelligence. Jia Zhangke interrogates AI vision models' capacity for narrative coherence, imagining a future where efficiency erases empathy. These artists expose algorithmic limitations while proposing embodied, pluralistic approaches.

The panel invites media scholars, art historians, artists, and AI researchers to address key questions: What do models mean to art or AI? How do artistic practices and AI relate through synergy, conflict, or mutual reinvention? How does art confront datasets, and what can each field learn from these clashes? In what ways can artistic practices reframe our understanding of intelligence itself? By probing these questions, the panel explores pathways beyond data-centric approaches to modeling intelligence.