ChatGPT is the language of the unconscious | Alenka Zupančič

Quick Overview

Alenka Zupančič argues that contemporary AI-generated content, exemplified by ChatGPT, functions like a distorted, hyperbolic echo of the unconscious, producing an "ironic" effect where the subject is both present and absent, ultimately failing to achieve true liberation or genuine thought, contrasting with Lacanian psychoanalysis.

Key Points: President Trump popularized a hyperbolic style of speaking that disregards precision, which followers understood figuratively, not literally. The irony in AI like ChatGPT is that it speaks literally, yet its output is figuratively divorced from true substance, resembling a caricature of 'gospel truth'. Lacanian psychoanalytic practice identifies moments of the real or the impossible, contrasting with the structural absence found in AI-generated discourse. The author argues that the subject produced by AI discourse is an effect of discourse, not a genuine subject, lacking a center or external grounding. The analysis critiques the New York Times article which suggested generative AI complements the 'Maga style' rhetoric, noting that this style itself is an effect of a 'discursive loop'. The inherent irony in Trump's style—telling the opposite of what he means—is contrasted with AI, which states things literally but means something else, creating an aesthetic of 'pure subjective unconsciousness'.

Context: Philosopher and psychoanalyst Alenka Zupančič discusses the nature of contemporary large language models (LLMs) like ChatGPT, contrasting their output with psychoanalytic concepts, particularly Jacques Lacan's notion of the Real and the subject. She examines how AI-generated language, despite its sophistication, reflects a structural lack or irony that prevents it from achieving genuine subjectivity or liberating insight.

Detailed Analysis

Alenka Zupančič critiques the contemporary reliance on AI language models like ChatGPT, framing their output as a hyperbolic, figurative echo of the unconscious, popularized by figures like Donald Trump who used hyperbole that his followers interpreted non-literally. Zupančič points out the irony that AI, while speaking literally, produces text that signifies figuratively, devoid of the 'real' that Lacanian analysis seeks. She contrasts this with the psychoanalytic concept of the subject, which emerges from a lack or an 'impossible point' within discourse. The AI's output, she suggests, is structured around an endless feedback loop of self-referentiality, lacking the external grounding of genuine subjectivity. She references a New York Times article analyzing Trump's style, noting that the AI-generated content mirrors this rhetorical approach—saying one thing but meaning another—but lacks the genuine 'political' or 'existential' weight. Zupančič concludes that the AI's output, while simulating structure, is fundamentally an effect of discourse that ultimately confines the subject within a loop of self-referentiality, rather than leading to liberation or genuine thought.