Generative AI has become integrated into students’ writing practices in higher education. While institutional discussions often focus on responsible use, transparency, and prompt design, a more fundamental pedagogical question remains underexplored: what intellectual and epistemic resources enable students to critically evaluate AI-generated suggestions rather than simply accept them? This presentation draws on an ongoing qualitative study examining reformulations generated by several AI writing tools from a corpus of academic texts in the humanities and social sciences. Preliminary findings reveal recurring tendencies across platforms, including the attenuation of critical markers, the simplification of complex arguments, the reduction of historical contextualization, and the standardization of discursive forms. Although such transformations may improve readability and fluency, they can alter dimensions that are central to a meaninfgful and creative scholarly writing. The presentation argues that the educational implications of these findings extend beyond teaching students how to use AI effectively and ethically. The capacity to recognize what may be lost through algorithmic reformulation depends not only on technical skills but also on familiarity with disciplinary traditions, exposure to deep thinking, diverse writing styles inculding litterary writing, and experience engaging with complex texts. From this perspective, reading becomes a crucial dimension of AI literacy. Students who have encountered a wide range of scholarly and non-scholarly voices are better positioned to adopt a more confident stance toward AI-generated revisions, questioning, modifying, or rejecting suggestions. The presentation concludes by proposing pedagogical approaches that cultivate interpretive judgment, stylistic awareness, and confidence in one’s own voice, positioning reading not as an alternative to AI literacy but as one of its essential foundations.
Banafsheh Karamifar is an Assistant Professor in Linguistics and French as a Second Language at Laurentian University, Canada, and an Adjunct Professor in the Integration of Technology in Education at the University of Ottawa. She has published on language and power, media discourse, plurilingualism, cultural diversity, and technology-enhanced education. She is co-editor of a two-volume edited collection on artificial intelligence and discourse, published by Springer in 2026. Her current research examines the discursive, pedagogical, and epistemic implications of artificial intelligence.