AI-Mediated Writing Instruction in Higher Education: A Systematic Review of Empirical Evidence

Abstract

The rapid integration of generative Artificial Intelligence (AI) into higher education writing instruction is outpacing pedagogical frameworks, creating profound disruptions in how writing is taught, assessed, and valued. While AI shifts writing from individual production to collaborative human–AI processes, instructors face escalating challenges, including the erosion of traditional authorship, uncertainty in evaluating AI-mediated work, threats to assessment validity, and growing student dependency on AI tools. These tensions expose a widening gap between technological adoption and pedagogical preparedness, placing faculty at the center of unresolved ethical, instructional, and institutional dilemmas. This systematic review synthesizes empirical research published between 2023 and 2025 on generative AI (e.g., ChatGPTand other GPT-based systems) in higher education writing instruction. Following PRISMA guidelines and SPIDER framework, 19 peer-reviewed studies were analyzed. Findings suggest that AI can enhance drafting, revision, and feedback processes, improving coherence, metacognition, and writing confidence. However, these benefits are accompanied by persistent concerns regarding ethical ambiguity, inconsistent policy guidance, and insufficient faculty training. The review contributes theoretically by reconceptualizing writing pedagogy for AI-mediated processes and practically by providing guidance for instructional design, assessment strategies, and institutional policy. Gaps identified include a lack of longitudinal studies, limited exploration of faculty perspectives, and inconsistent integration of AI literacy and ethical considerations. Implications for research, practice, and policy are discussed.