Exhibition views and video work
PUBLICATION
The exhibition catalogue complements the spatial installation by making our working process visible. It brings together “observation units” and collages that accompany the development of generative AI in relation to the representation of speech. Using the same prompt – “close-up speaking mouth” – we repeatedly queried different models and their versions between April 2024 and August 2025. The excerpts document processes of recording, analysing, and questioning these synthetic image spaces and mark an interim state within an ongoing writing process embedded in the accelerated evolution of generative AI.
Insight into our working process:
translating video-to-concrete
The multi-part concrete relief preserves an early stage of video generation, a moment when lip movements were still incoherent. It is based on a six-second screen recording of a generated, speaking mouth. Using custom-built software, the 150 frames of the video were translated into a single 3D object, which was 3D printed and cast in concrete.
Within this imprint, not only the early phase of AI video generators become visible, but also the traces of their interfaces: buttons, company names, model identifiers. Frozen in its movement, the video becomes a fossil, while the static elements—the tools of the interface—are stripped of their function and interfering with the motif.
