Sudden Occlusion is "a phantasmagoric horror-road movie" directed by Rebecca Shapass and "set in the simulation of an autonomous car’s AI-training system." The project, experienced as an immersive 360° VR video, reconstructs Pittsburgh from the open-source Aurora Multi-Sensor point-cloud Dataset. The work places the viewer inside a fragmented digital city, where, through embodied navigation, visibility is partial and spatial continuity is deliberately disrupted.

As the car (the viewer) attempts to recollect the post-industrial landscape of Pittsburgh, the sample training model it is able to access renders its limited memory full of fissures. Rather than arriving at any one place, a journey through this model’s simulation acquaints viewers with the machine’s method of perception and poses for them a question of identifying / disidentifying with the ways the AI forms knowledge and holds memories of the world. By emphasizing occlusion, compression, and absence, the work ultimately questions what is lost when cities are experienced as stitched datasets rather than lived environments.