Artificial Analysis: Taking Chances with the Imaginary

Abstract

Recent advances in large language models (LLMs) have intensified speculation about the possibility of artificial agents functioning as therapeutic listeners. This paper approaches that question from a psychoanalytic perspective by placing contemporary AI within the conceptual lineage of cybernetics, probability theory, and Freud’s early metapsychology, as interpreted by Jacques Lacan in Seminar II. Drawing on Lacan’s analysis, a detailed comparison is developed between specific functions in LLMs and Freud’s early accounts of the psychical apparatus, particularly those in the Project for a Scientific Psychology. While these structural affinities reveal striking parallels between LLMs and the psychoanalytic subject, the paper contends that such similarities do not suffice for the subjective transformations necessary to promote the capacity for analytic listening. The paper concludes that the limitation of artificial intelligence as an analytic listener does not stem from its machinic nature, but from its exclusion from the analytic dialectic itself. As such, current LLMs function therapeutically at a level comparable to any “approximately normal person,” as Freud put it, illuminating both the specificity of psychoanalytic listening and the stakes of its distinction in an era of artificial interlocutors Freud, (1912/1958, p. 116).

How to Cite: