The Uncanny Valley: AI, Transference, and the Artificial Other

The uncanny valley is more than an engineering problem or a temporary lapse in comfort when interacting with human-like machines. Drawing on empirical work on perceptual mismatch, prediction violation, and category conflict, as well as revisiting Freud’s formulation of the uncanny through Solms’ revised translation, uncanny reactions can be seen as clinically meaningful affects that signal the collision of familiarity and strangeness in a way that invites unconscious fantasy, defensive operations, and object-relational organization. A psychoanalytic account of the uncanny valley as an “uncanny object” represents an almost-person that is close enough to recruit attachment and identification, yet different enough to become a privileged surface for projection, splitting, and projective identification. In this frame, AI agents and chatbots can become potent transference objects, sometimes idealized as perfectly responsive and sometimes experienced as cold, intrusive, or lifeless, with rapid oscillations between these positions. This is illustrated by a psychodynamic case involving an AI chatbot that demonstrates how such attachments enter the analytic situation and shape alliance, authority, and disappointment. Finally, the uncanny experience is related to tensions around aliveness and deadness, including the life and death instincts and masochistic compromise formations.