Abstract "Automata" is a research-creation project that examines how the delegation of interpretation and creative agency to intelligent machines transforms the conditions of human freedom and judgment. Rather than asking whether machines “think,” the project investigates a structural shift: the relocation of interpretative authority from visible human construction to opaque algorithmic inference.

Through a reflexive documentary centered on Swiss automata-maker François Junod, internationally recognized for his mechanical automata, the film contrasts two regimes of intelligibility. On one side, a historically sedimented conceptual world grounded in visible causality, repairability, and embodied expertise; on the other, contemporary generative AI systems operating through large-scale statistical architectures whose internal processes remain inaccessible to users.

By integrating a generative language model directly into the filmmaking process—inviting it to produce its own portrait of the protagonist, to propose alternative framings, and at times to narrate from a machinic perspective—the project stages an experimental confrontation between human and algorithmic interpretation within the cinematic apparatus itself.

Engaging the SAR 2026 theme of endangered conceptual worlds, "Automata" treats embodied mechanical intelligence as an epistemic counter-model to algorithmic flattening, while testing whether materially grounded expertise can remain a viable locus of meaning when confronted by a persuasive, and potentially rival, algorithmic imagination.