Martin Thellefsen, Amalia Nurma Dewi, Bent Sørensen
Purpose This paper aims to explore prompts and prompting in large language models (LLMs) as dynamic semiotic phenomena, reconceptualizing the practice from a technical input mechanism to a communicative and epistemic act that involves iterative sign formation, interpretation and refinement. Design/methodology/approach The theoretical foundation rests on Peirce's semiotics, utilizing the triadic model of signs (representamen, object, interpretant), his nine sign types and the Dynacom model of communication. Analytically, the LLM is positioned as a semiotic resource generating interpretants in response to user prompts to participate in meaning-making within a shared universe of discourse. Findings Prompting constitutes a semiotic and communicative process that redefines how knowledge is organized, searched, interpreted and co-constructed in digital environments. Originality/value This perspective invites a reimagination of the philosophical foundations of meaning, interpretation and epistemic agency in an age defined by human–machine communication and computational semiosis.