Does generative AI improve university students’ independent English language proficiency?

The growing integration of generative artificial intelligence (GenAI) into higher education raises important questions about its contribution to students’ independent learning outcomes. This study examines whether structured AI-assisted learning improves university students’ independent English language proficiency compared with unrestricted AI use and conventional instruction. A quantitative, quasi-experimental design is proposed involving 150 undergraduate and postgraduate students assigned to three instructional groups. English proficiency is assessed through pre-tests, post-tests, and delayed tests conducted without AI assistance, covering grammar, vocabulary, academic writing, reading comprehension, and speaking. Regression analysis is proposed to examine the roles of instructional approach, self-regulated learning, prior AI experience, baseline proficiency, and age in explaining proficiency gains. The study seeks to distinguish AI-supported task performance from independently demonstrated language competence and retained learning. Its anticipated contribution is to inform evidence-based integration of GenAI into university language education, emphasising structured use, learner autonomy, and independent assessment.