Operational ecologies—how does a machine see?
Talisa Rubina
Royal College of Art
Drawing on safety-enforcing surveillance systems used in spaces such as train stations, the installation explores how machines interpret, classify and monitor human behaviour. Using a live camera input, driven through real-time computational analysis, visitors become both participants and subjects of observation. Four sculptural screens reveal separate layers of this process (object detection, spatial measurement, behavioural tracking and scene classification) while a large-scale projection combines them into an algorithmic assessment of risk. The projected image shifts between clarity and pixelation according to the system’s confidence in its classifications. By making these usually hidden processes visible, the work explores how automated systems construct notions of safe and unsafe behaviour, inviting reflection on the growing authority of algorithmic forms of seeing and judgement. An earlier iteration of the work was exhibited at Orleans House Gallery in London.