AI-Generated Writing Feedback for Critical Thinking Development in EFL Writing: A Systematic Literature Review

AI-generated writing feedback has become a common feature of English as a Foreign Language (EFL) writing instruction, offering timely support during drafting and revision. This systematic literature review examines AI-mediated feedback on learners’ written texts, including automated writing evaluation (AWE), AI writing assistants, and generative AI or large-language-model (LLM) tools when they produce evaluative, corrective, explanatory, or dialogic feedback on a learner’s draft. It does not review AI used only to generate text, score writing, or provide general language practice without feedback on writing. Nineteen peer-reviewed empirical studies published between 2022 and 2026 were synthesised thematically. The review reports patterns in the available evidence rather than an overall effect size. Direct measures of critical thinking (CT) were distinguished from writing outcomes that may serve only as CT-related proxies, such as coherence, organisation, and revision quality. The evidence suggests that AI-mediated feedback may support argument development, evidence use, evaluative revision, and reflection when it is embedded in structured instruction. Evidence for sustained reasoning, independent judgment, and self-regulated evaluation of AI suggestions remains limited and heterogeneous. AI-mediated feedback should therefore be treated as a conditional pedagogical resource, not as an automatic means of developing critical thinking.