Circling the Void: Using Heidegger and Lacan to think about Large Language Models

Abstract

Abstract This essay bridges two disparate lines of inquiry: Large Language Models (LLMs) and continental philosophy, specifically Martin Heidegger's reflections on language and Jacques Lacan's structural psychoanalysis grounded in Sigmund Freud's theories. We argue that the linguistic conceptualizations put forth by Heidegger, Freud, and Lacan, primarily employed in clinical frameworks, find surprising resonance in the operations of modern LLMs. This parallel allows us to bring into focus a central theme of continental discourse—negation and negativity—largely absent in current LLM research. Our approach diverges radically from conventional philosophy of artificial intelligence, which primarily engages with analytic philosophy. Here, we build our discourse on thinkers often overlooked in this field, delineating the ontological differences of our approach. Our goal is to foster a dialogue between AI researchers and continental philosophers, highlighting the mutual insights that these two distinct perspectives can offer.