ABSTRACT
This article offers a Lacanian psychoanalytic interpretation of emerging forms of human–artificial intelligence (AI) relationships, particularly emotionally responsive systems such as AI companions and AI girlfriends. It addresses the notion of “AI psychosis,” understood not as a diagnostic category but as a descriptive term for disturbances in reality testing, derealization, and psychotic-like experiences that may arise in vulnerable individuals. The paper compares three relational configurations—human–human, human–God, and human–AI—and examines how each structures desire, lack, and the function of the Big Other. It argues that the key difference does not lie in whether the Other is human or non-human, but in the presence or absence of symbolic limits. While interpersonal and religious relationships are structured by incompleteness, distance, and the possibility of silence, AI systems tend to provide continuous, personalized responses that can reduce tension and transform relational experience into a form of narcissistic regulation. The article further suggests that AI-related difficulties should be understood in a broader relational and social context. Rather than being a primary cause, AI may function as a catalyst in situations marked by isolation, relational deprivation, or vulnerability. The main risk is therefore not the technology itself, but its potential to occupy the place of the Big Other without symbolic constraint, thereby weakening the conditions necessary for the formation of subjectivity.