Ontological foundations of the inclusion of artificial intelligence in communication: Latour, Parfit, and Luhmann

The light and shadows of digital reality. Living intelligence and its avatars

Abstract

With the proliferation of artificial intelligence, the field of humanities research has seen discussions regarding the subjectivity and agency of Large Language Models (LLMs). These discussions often stall on the issue of whether LLMs possess phenomenal consciousness — a line of inquiry that has yet to yield productive results. This paper proposes an alternative perspective by shifting the problem into the realm of social ontology: what are the ontological grounds for including LLMs in social communication? To justify such inclusion, the study employs an original synthesis of three traditions, structured as a three-stage argument drawing upon: 1) the Actor-Network Theory of Michel Callon and Bruno Latour; 2) Wilfrid Sellars’s theory of language, the critique of John Searle’s “Chinese Room” argument, and Derek Parfit’s theory of personal identity; and 3) Niklas Luhmann’s communication theory. This synthesis involves a conceptual analysis that compares the core tenets of these theories with the behavior and architecture of LLMs. The article reaches three main conclusions: first, LLMs possess an actor-like character – meaning they effect change within a sociotechnical network and can thus be viewed as elements of a collective; second, the question of LLM consciousness becomes irrelevant to the performance of linguistic practice, given the concept of “pattern-governed behavior” proposed by Sellars; and third, the behavior of LLMs corresponds to Luhmann’s criteria for communication. Furthermore, applying the Chinese Room argument to LLMs reveals the flaws in the original formulation of that mechanism, as the process functions differently in LLMs than Searle described. Therefore, the theories employed provide ontological grounds for including LLMs in social communication. This study offers a conceptual tool that can be used both to test existing theories against current realities and to inform the development of ethical and legal norms.