ARTIFICIAL INTELLIGENCE IN ENGLISH LANGUAGE TEACHING IN PAKISTAN AND GLOBAL EFL CONTEXTS (2020–2025): A COMPARATIVE SYSTEMATIC REVIEW

Abstract

This systematic review focuses on the incorporation of artificial intelligence (AI) in English language teaching and learning in 2020-25 and is comparatively discussed with respect to Pakistan and the rest of the world. This was narrowed down a bit to Pakistan vs rest of the world to Pakistan and global EFL contexts since the Pakistani evidence is still in its early stages and cannot be considered an entirely symmetrical evidence base as compared to the larger international body of evidence. The search was conducted by PRISMA 2020 protocols and organized according to the SPIDER model in Scopus, Web of Science, ERIC, Semantic Scholar, Google Scholar, and publisher databases and then by backward and forward citation chasing. The overall synthesis comprised 29 studies: seven studies about Pakistan and 22 studies in the global EFL/ESL setting. The synthesis demonstrates that the integration of AI in ELT is prevailed by generative AI and writing feedback via automated writing assistants, especially ChatGPT, Grammarly, and other writing-help applications. Chatbots to practice conversation, AI speaking partners, automatic speech-recognition systems like ELSA Speak, adaptive learning systems, teacher readiness and academic-integrity responses are other areas. Mixed-method and intervention studies, as well as quasi-experimental studies, are more likely to be found in the global evidence base; Pakistan-based research is focused on the perceptions and attitudes of learners and teachers and on early exploratory designs, with one more powerful quasi-experimental study on ChatGPT-based argumentative writing. In most contexts, AI tools were typically reported to contribute to the immediacy of feedback, practice opportunities, learner confidence, vocabulary/writing support, and reduction of teacher workload. Nevertheless, various challenges were recurrent, which were hallucinated or inaccurate feedback, over-reliance, unequal access, ineffective teacher training, risks of academic integrity, and lack of longitudinal classroom evidence. The review ends with a proposal of a context-sensitive model of AI integration in Pakistani ELT that integrates AI literacy, teacher mediation, progressive classroom application, ethical evaluation design, and empirical evaluation of outcomes.