Priors is a 16-minute generative video installation that investigates prediction as both a cognitive process and a computational mechanism in artificial intelligence. The title refers to Bayesian inference, where "priors" are the probability distributions that shape how new information gets interpreted—the assumptions the brain carries forward to interpret sensory data, and the embedded expectations that guide predictive models in machine learning.
Built from custom AI workflows, the work stages a sequence of shifting environments where forms seem to anticipate themselves, unfolding in ways that blur the line between emergence and recall. These two domains—human cognition and machine prediction—collapse into one perceptual field: landscapes, patterns, and fragments of motion appear to arise from the viewer's own act of looking, as if perception itself were shaping the terrain.
Rather than presenting fixed images, the work unfolds as a continuous flow of probabilities and transformations. The viewer is suspended in a state of partial recognition, where prediction and memory intertwine. Priors reflects on how both human and machine vision are structured not by what is present, but by what is expected.
What does it mean to see in an age when machines are learning to predict the world alongside us? And what might we discover about our own vision by watching theirs unfold?
Built from custom AI workflows, the work stages a sequence of shifting environments where forms seem to anticipate themselves, unfolding in ways that blur the line between emergence and recall. These two domains—human cognition and machine prediction—collapse into one perceptual field: landscapes, patterns, and fragments of motion appear to arise from the viewer's own act of looking, as if perception itself were shaping the terrain.
Rather than presenting fixed images, the work unfolds as a continuous flow of probabilities and transformations. The viewer is suspended in a state of partial recognition, where prediction and memory intertwine. Priors reflects on how both human and machine vision are structured not by what is present, but by what is expected.
What does it mean to see in an age when machines are learning to predict the world alongside us? And what might we discover about our own vision by watching theirs unfold?
16 minutes, 4K, custom dataset, generative AI, single-channel display
Previously exhibited:
2025 at Phillips London, Digital Art Awards
2025, Digital Art Awards with HOFA Gallery
2025 at Phillips London, Digital Art Awards
2025, Digital Art Awards with HOFA Gallery