Rethinking AI in Writing Pedagogy: From Fluency to Rhetorical Competence

Abstract Generative AI tools such as ChatGPT are reshaping academic writing in higher education, offering students unprecedented fluency, coherence, and accessibility. Yet these seductive characteristics, together with serious rhetorical limitations, raise concerns about authorship, integrity, and the development of critical thinking. The similarities and differences between AI generated texts and human writing, and the implications of these for writing instruction, however, remain underexplored. This study aims to bridge this gap by synthesizing five studies conducted by Jiang and Hyland, which systematically compare ChatGPT--generated essays with those authored by British undergraduates. The studies reveal a consistent pattern: while ChatGPT excels in producing structurally polished texts, it underperforms in rhetorical depth, interpersonal engagement, and authorial presence. Building on these findings, the paper proposes a framework for integrating AI into writing pedagogy. This positions ChatGPT as a supportive tutor rather than a proxy author, stressing both its advantages and limitations. It offers practical pathways for educators and underscores the importance of cultivating critical AI literacy, enabling students to navigate new writing contexts where human creativity and algorithmic fluency intersect.