Yuda Sakti Susena, Fajar Arianto, Khusnul Khotimah, Rico Eko Andrianto, Muchammad Syuhada’
Generative and conversational artificial intelligence (AI) has rapidly become embedded in everyday educational practices, including activities closely related to reading, text comprehension, and academic literacy. However, existing review literature remains conceptually fragmented, particularly regarding AI’s epistemic, dialogic, affective, and moral implications for learning. This systematic review synthesizes evidence from 80 review-level studies published between 2018 and 2025 to examine how generative AI functions as an epistemic authority, cognitive scaffold, dialogic participant, affective interlocutor, and moral reference point within educational contexts relevant to reading and literacy practices. Following PRISMA 2020 guidelines, systematic reviews, scoping reviews, and meta-analyses were analyzed using thematic synthesis. Findings indicate that generative AI can support critical reading, reflective text engagement, and dialogic interpretation when integrated through intentional pedagogical design. Conversely, uncritical use is associated with automation bias, cognitive offloading, homogenized discourse, and attenuated evaluative judgment. These outcomes are shaped less by technology itself than by instructional design, assessment practices, and institutional governance. © 2026 Taylor & Francis Group, LLC.
Department of Educational Technology, Universitas Negeri Surabaya, Surabaya, Indonesia; Department of Guidance and Counseling, Universitas Negeri Surabaya, Surabaya, Indonesia
