This article proposes a rhetorical theory of generative AI usage based on the concept of “double articulation” (Deleuze and Guattari 2005). Amidst the widespread diffusion of Large Language Models (LLMs), rhetorical theory faces a dual crisis of utility and hermeneutics: users struggle to validate the epistemic worth of AI-generated text and to interpret its distributed authorship. Moving beyond the binary of AI as either a neutral tool or an autonomous agent, this article argues that text production with LLMs is a process of reciprocal stratification where human intention (inventio) and machinic probability (stochastic generation) are inextricably entangled. By tracing a genealogy of rhetorical generativity – from the topoi of Aristotle and the ars combinatoria of Ramon Llull to the humanist ideal of copia – the study reveals that generative AI does not constitute a rupture but an intensification of rhetorical techné. The article analyzes how core rhetorical structures (topoi, doxa, imitatio, copia) are operationalized in algorithmic environments and argues that successful communication depends on a “Rhetorical AI Literacy.” This literacy recenters the human user not as a mere prompter, but as a critical editor exercising judgment (iudicium) to transform machinic probability into situational appropriateness (aptum). Ultimately, the article demonstrates that the “human-in-the-loop” is not a temporary fix, but the constitutive horizon of meaning in an age of hybrid textuality.
Markus Gottschling (Sat,) studied this question.