Presentation
Uchronie (uchronia) is an interactive installation that uses artificial intelligence to generate dystopic stories and fabulatory portraits inspired by objects personally significant to each interactor.
When users approach the installation, they can press an old intercom’s button to verbally identify an object of importance and explain why it matters to them. Leveraging this input and a surreptitiously captured image of the user, the system invents and recites a speculative and dystopic narrative about the user’s future with that object. At the conclusion, a surreal AI-generated portrait, inspired by the interaction, is displayed. The last thirteen of those portraits float above the installation on the screen of obsolete iPhones. Their clichéd, distinctly AI aesthetics, highlight shortcomings of current generative models while foreshadowing the radical societal changes that are forthcoming.
Ironically, the agent that produces both story and portrait is forced into an introspection that is very self-critical, often describing how ill-fated our reliance on technology will be. Echoing Kodak’s 19th-century promise — “You Press the Button, We Do the Rest” — Uchronie reveals that, just as photography’s objectivity was an illusion, these AI-generated outputs are likewise tainted by all the biases embedded in generative systems.
User-feedback emerges through two CRT television sets which intentionally distort the clean digital output and highlight how technologies always filter, frame and reshape content. This mediarcheological approach explores our broken relation with outdated technological materialities. In parallel, a physical web of wires underscores our entanglement with the material world and how the “digital” always remains bound to the tangible.
Video
Photo Gallery
Tech Rider
The technical rider should answer most questions regarding the physical, electrical and technical needs for this installation.
Localization
The interface language can easily be updated to match local language. The audio stories are told in the same language the user spoke in (provided it is one of OpenAI’s supported languages).
Acknowledgements
I would like to thank the following partners for their support: