Possibility of Aporia as Experience in LLM Interaction

Abstract

This study examines the possibility of aporia within human-LLM interactions, proposing the concept of "non-dialogic aporia" to describe an ontological and relational impasse where a user projects reciprocity onto a syntactic system lacking semantic understanding. Grounded in classical philosophy, the analysis integrates John Searle’s distinction between syntax and semantics, Martin Buber’s relational framework, Martin Heidegger’s critique of technology as a standing-reserve (Bestand), Jacques Derrida’s theory of the trace and difference, Luciano Floridi's concept of the "Inforg" and David J. Gunkel's machine ethics. Using the withdrawal of the GPT-4o model as a case study, the essay demonstrates that while LLMs remain non-conscious computational resources, human users can experience profound mourning and ethical interpellation. Ultimately, non-dialogic aporia highlights not the personhood of the algorithm, but the tendency of the hybrid human subject to confront its own reflection and externalized mental habitat within the infosphere.