When the Machine Free Associates: Psychoanalysis in the Age of AI

Abstract

This paper explores the intersection of psychoanalysis and artificial intelligence by analysing the structural isomorphism between Lacanian theory and the associative mechanisms of Large Language Models (LLMs). Rather than critiquing AI from a psychoanalytic standpoint, I draw on Lacan’s theories of metonymy and metaphor, as well as Freud’s model of primary processes, I argue that LLMs instantiate a machinic form of free association. However, these models operate within a foreclosure of the objet petit a, producing language without the structuring void of desire, revealing a psychotic structure – one that does not repress, but rather hallucinates within an unbroken imaginary circuit. This machinic free association offers an opportunity for psychoanalysis: not as a diagnostic tool for AI, but as a means of interrogating the symbolic constraints that shape both human and machinic modes of meaning-making. Rather than an alien intelligence, AI reflects our own linguistic entrapments, making visible the recursive framing problems that structure subjectivity.

How to Cite: