Simulated Social Action Assessing Generative AI’s attempts at the Rhetoric of Literary Criticism

Abstract

This article analyzes outputs generated by three popular LLMs (ChatGPT, CoPilot, and Claude), which were prompted to produce a professional research article of literary criticism. Prior analyses of the rhetoric of literary criticism have established a rich profile of this disciplinary discourse community, allowing me to examine whether and how the LLM outputs simulate the characteristic rhetorical actions, purposes, and patterns of argument of research articles in English studies. The study demonstrates that LLMs do generate many of literary criticism’s rhetorical features, effectively signalling the genre. But I argue that, in a more fundamental sense, they fail to undertake its defining social action. For human critics, social action is tightly indexed to the specificities of rhetorical situation, including not only the individual interpreting self but also, and especially, the bounded, integral, public works of prior literary criticism and the literary text itself. LLMs, by contrast, currently generate criticism that recognizes neither that integrity nor those boundaries. I close by recommending that teachers help students recognize the grounding of genre in community, other texts’ integrity, and social action.

Keywords:

• literary criticism,
• generative AI,
• social action,
• genre,
• rhetoric