Large language models and M. Heidegger’s philosophy of language

Archive of Scientific Articles

ISSUE:

Manuscript. 2025. Volume 18. Issue 4

COLLECTION:

Ontology and Epistemology

All issues

Abstract. The article provides a philosophical analysis of the ability of large language models (LLMs) to generate authentic meaning. The aim of the study is to identify the fundamental ontological conditions for meaningful discourse based on Martin Heidegger’s philosophy of language and to apply them for a critical assessment of LLM-generated texts. The article compares key concepts of M. Heidegger’s philosophy (“Dasein”, “In-der-Welt-sein”, “Sorge”, “Sein-zum-Tode”) with the architecture and principles of operation of LLMs. The scientific novelty of the research lies in a new approach to the philosophical analysis of large language models through the prism of M. Heidegger’s existential-ontological philosophy. Unlike the epistemological and cognitivist interpretations dominant in contemporary literature, which focus on questions of information processing and functional equivalence to human intelligence, this study transfers the problematic to a fundamental ontological level, revealing the structural conditions for the possibility of authentic meaning-generation. The innovative aspect is establishing a direct connection between the absence of existential structure in LLMs and their inability to generate authentic meaning, which shifts the discussion of machine understanding from the plane of increasing computational power and data volumes to the realm of fundamental questions about the nature of human being and language. The research makes an original contribution to understanding LLMs as a technological embodiment of Gestell, demonstrating how contemporary language models institutionalize an ultimately technical relation to language, transforming it from the “house of Being” into a calculable resource, which has far-reaching philosophical and cultural consequences for understanding the nature of communication and cognition in the digital age. The results amount to the following: LLMs operate outside the existential structure of Dasein (human presence), reducing language to a calculable statistical resource (Bestand) and stripping it of its event-of-truth dimension. The technological paradigm (Gestell) underlying LLMs engenders epistemic, ontological, and anthropological risks, including the spread of pseudo-knowledge and the devaluation of human thought.