<![CDATA[The diffusion of generative artificial intelligence (AI) has challenged assumptions about how English L2 writing is taught, supported, and assessed. This systematic literature review synthesised peer-reviewed empirical evidence on generative AI in English L2 writing, focusing on pedagogical transformation, feedback and assessment, and academic literacy. Following PRISMA 2020, searches of Scopus, PubMed, and ScienceDirect identified 912 records. After duplicate removal and eligibility screening, 18 empirical studies published between 2023 and 2025 were included, although the search window covered 2020-2025. The eligible studies were appraised with the Mixed Methods Appraisal Tool (MMAT 2018), with 14 meeting at least 4 of 5 criteria under the manuscript's descriptive appraisal rubric and four scoring 3.0-3.5; none was excluded solely on quality. Owing to substantial heterogeneity in designs and outcomes, findings were synthesised narratively and no quantitative meta-analysis was attempted. Evidence indicates that generative AI can improve surface-level accuracy, revision activity, and selected language or self-regulatory outcomes, while effects on higher-order rhetorical and critical competencies remain inconsistent. AI is most defensible as a teacher-guided complement, with continuing concerns about academic integrity, over-reliance, assessment validity, and equity. The review identifies priorities for longitudinal, multi-model, and multi-site research.]]>
Copyrights © 2025
