The Impact of AI-Assisted Learning on Non-Native Graduate Students in Overseas Study Programs: A Qualitative Analysis of Learning Anxiety, Classroom Interaction, and Knowledge Construction

This study, grounded in Second Language Anxiety Theory and Constructivist Learning Theory, investigates how artificial intelligence-assisted tools, primarily ChatGPT, shaped non-native graduate students’ learning anxiety, classroom participation, knowledge construction, and interaction during a five-week short-term study program on the U.S. West Coast. Participants were master’s and doctoral students from three national universities in Taiwan. A qualitative design was adopted, employing participant observation and semi-structured interviews, and the data were analyzed using a three-stage Grounded Theory coding process. Findings indicate that participants initially reported high levels of anxiety due to unfamiliar accents, rapid speech, and limited linguistic confidence. With the integration of ChatGPT, participants reported reduced anxiety as they were better able to comprehend instructional content, clarify terminology, and formulate appropriate responses. Observations and interview data further suggested increased classroom participation, as AI support enabled learners to move from passive reception to more active questioning, discussion, and knowledge construction. Although AI mediation appeared to alter some face-to-face interaction patterns, participants generally demonstrated greater engagement in classroom learning. Overall, the results suggest that AI-assisted tools can provide meaningful cognitive and emotional scaffolding for non-native learners in overseas academic settings, supporting learning effectiveness, self-efficacy, and engagement. The study offers practical implications for the design of future short-term study programs and AI-supported instructional practices.