The Models is an interactive installation by artist duo dmstfctn exploring the improvisational nature of generative AI, and its sometimes nonsensical, sycophantic, or deceptive character. A live simulation running continuously for a year renders improvised theatrical scenes with a style and cast inspired by Italian Commedia dell’Arte theatre. Dialogues are generated by a system of AI models prompted to impersonate traditionally servile, lying or antagonistic Commedia masks – Arlecchino, Balanzone, Brighella, Colombina, Pantalone and Pulcinella, improvising around a set of controversial props that symbolise superstition, folklore, or falsehoods.
Audiences use their phone to select masks and props for each scene in response to a backdrop selected at random – contributing to the creation of a textual prompt that is turned into dialogues by a system of AI models working in the background. Influenced by the audiences’ prompt and by Commedia archetypes present in their training dataset, the models’ output is unpredictable and reliably rich in confabulation, mistakes, and surprise. Once a scene begins, audiences can throw coins and flowers, or tomatoes and eggs for feedback, in keeping with traditional audience participation during Commedia dell’Arte scenes.
About dmstfctn
Artist duo dmstfctn explore complex systems through the lenses of simulation, fiction and experience. Their immersive performances, video installations and video games demystify systems by replicating and exploring them together with audiences, and ‘re-mystify’ them by building new stories and worlds atop them. Since 2018, they have performed and exhibited at Serpentine, Berghain, HWK, LUMA, SOLO and festivals such as Unsound, CTM, and transmediale. Their work around AI anomalies saw them collaborate with scientific institutions such as The Alan Turing Institute and Europe’s top Supercomputer Leonardo. Since 2025, they are n-Space fellows at Somerset House Studios.