Research Program Overview — Discursive Signatures and Uses of Large Language Models

This record provides an overview of the research program Discursive Signatures and Uses of Large Language Models (LLM). The program investigates how different large language models produce self-referential discourse and how these discursive patterns influence their practical use in knowledge activities. Three recurrent discursive regimes are identified: functional translation (the model presents itself as a technical system) reflexive absorption (the model comments on its own operation) conversational dissolution (the model stabilizes a social interaction) The objective is not to attribute cognition or subjectivity to LLMs, but to analyze the operational consequences of these discursive signatures for research, documentation, and learning practices. This document is a research overview and methodological entry point to the associated working papers.