VietTESOL International Convention 2026

Advances in Artificial Intelligence (AI) have transformed tertiary education, yet the long-term impact of Generative AI (Gen-AI) on autonomous academic vocabulary acquisition in English-Medium Instruction (EMI) remains unclear. Therefore, grounded in the self-directed learning framework, this study aimed to investigate the perceptions of Vietnamese EMI undergraduates regarding the influence of Gen-AI on learning discipline-specific terms. A group of 120 sophomore non-English major students, who were enrolling in EMI programs at a private university in Ho Chi Minh City, was recruited using a convenience sampling method. Utilizing a mixed-methods design, a closed-ended questionnaire was conducted to collect data, and 10 selected students participated in semi-structured interviews to gain deeper qualitative insights. The collected quantitative and qualitative data were subsequently analyzed using descriptive statistics (i.e., frequencies and percentages) and content analysis, respectively. The findings indicated that students perceived gen-AI as most beneficial for better comprehension of technical terms, generating contextualized examples in assignments. The students also raised their concerns about inaccurate definitions, over-reliance, and inconsistent verification practices. The study offered pedagogical implications, suggesting a multistage AI integration framework to scaffold learning. Specifically, instructors can apply these findings by designing tasks that require mandatory verification of AI-generated terms against academic corpora and utilizing AI as a non-threatening environment for pre-class preparation to enhance student confidence and autonomy.