Mario Klingemann is an artist and describes himself as a skeptic with a curious mind. He taught himself programming in the early 1980s and has been creating algorithms that are able to surprise and to show almost autonomous creative behaviour ever since.
In this Keynote, he introduces his latest work, Interstitial Space 2019, which creates an open feedback loop between a portrait-generating adversarial neural network and the audience. Through a camera the system observes in real-time its own output that is being projected on to the walls and tries to identify facial features. These features get reinterpreted by a GAN (generative adversarial network) that has been trained on portraiture and become part of new output.
The noise and misinterpretation introduced by the neural models involved in this process cause the system to never repeat itself but at the same time reveal the nature of the data it has been trained on. By stepping into the gap between the camera and the projection, the spectator becomes part of the cycle and can attempt to take control over the emergent visuals.