English as a medium of instruction in the age of AI: Challenges, opportunities, and linguistic implications

This study aims to explore the role of AI in transforming EMI teaching and learning, the opportunities and challenges this brings, and the linguistic implications. The study has qualitative characteristics (thematic analysis) and uses secondary and documentary data, including peer-reviewed EMI and AI-in-education literature and sources, institutional literature and publications, policy guidance, and literature from discussions of higher-education practice. Sources have been coded iteratively and synthesised into nine analytical domains. AI provides access to vocabulary, grammar, translation, pronunciation practice, and immediate feedback, enables differentiated instruction, and frees up teachers' time to focus on instructional design and/or critical mediation. But benefits depend on pedagogical purpose. Without careful use, it can facilitate dependency and reduced interaction, lower verification, undermine assessment, and worsen the digital divide for students. Practices, expectations, and teacher readiness; student AI proficiency; ethical guidelines; digital access; and the commitment to maintaining the integrity of human interaction are all essential to linguistic outcomes. The article proposes a 5-dimensional framework for the responsible use of AI in EMI. It suggests that AI is not meant to replace the linguistic, pedagogical, and intellectual involvement of the human factor during the mediation action.