The AI Dilemma in EFL Classrooms at Higher Education

Generative Artificial Intelligence (AI) has rapidly reshaped English as a Foreign Language (EFL) learning in higher education, creating a paradox in which task efficiency coexists with cognitive passivity. This issue is rarely examined among English Education students, who are simultaneously language learners and future language teachers. This study investigates how these students navigate the dilemma of using AI in language classrooms, mapping both the opportunities for task efficiency and the challenges of cognitive passivity. An explanatory sequential mixed-methods design was applied at the English Education Department of Universitas Negeri Makassar. A Likert-scale questionnaire was distributed to 67 undergraduate students, followed by semi-structured interviews with 7 purposively selected students; quantitative data were analysed using descriptive statistics and the interview transcripts using thematic analysis. The findings reveal that students overwhelmingly value AI as an efficient virtual assistant for finding vocabulary (97.1%), correcting grammar (95.6%), and saving time (91.1%). At the same time, a notable minority report cognitive risks such as overreliance, dependency for idea generation, and reduced confidence when working without AI. Crucially, most students are not passive: they apply self-regulated learning strategies, setting strict personal boundaries (85.1%), critically analysing why AI makes corrections (86.5%), and deliberately editing AI output to preserve their authentic human voice (86.6%). The study contributes empirical evidence on how future language educators balance automated assistance with independent critical thinking, and offers practical recommendations for cultivating self-regulated AI use in EFL settings.