Looking at eyes that never looked back: photographic portraits, synthetic images and (mis)recognition

The photographed face has long been a locus of recognition, memory, and relation. In the portrait, we observe the trace of an encounter shaped by the unequal dynamics between photographer, subject, and viewer, and with it, the portrait’s connection to reality, time, and death. But what happens when the portrait is AI-generated and the human face no longer bears the marks of presence when someone appears before a lens? Through experiments with generative imaging, we examine how algorithmic processes reconfigure the photographic portrait. Drawing on Barthes’s ‘I am looking at eyes that looked at the Emperor,’ we consider the paradoxes that arise when photographic registration gives way to algorithmic aggregation. We trace the ethical and political stakes of this shift to Barthes’s temporal trace of a past encounter, Ariella Azoulay’s view of photography as a civil contract that unfolds through time and insists on ongoing responsibility, and to the loss of any encounter at all in synthetically generated portraits. Building on critical discourses of photography, facial recognition datasets and synthetic media, we consider how synthetic portraits appear to sever photography’s relational structure, while exposing and intensifying the inequities embedded within it.