EFL Learner Perceptions and Engagement of a Customized AI-led Class

Abstract

Technological advancement has enabled language educators to employ AI virtual humans as online instructors by customizing their characteristics, such as English varieties, to meet learners’ needs and preferences. As AI instructors become a viable option in classrooms, how they affect language learners’ learning warrants investigation. Building upon social presence theory regarding interpersonal relationships in an online environment, this study aimed to examine the role of social presence and AI instructors’ credibility in fostering learner engagement. Additionally, it examined the effects of variables within instructors on instructor credibility and learner engagement. In the study, a 2 (human or AI) x 2 (native or non-native English-speaking teacher) between-subjects design was utilized in an online experiment with 120 English learners. Regression and mediation analyses revealed, in AI-led classes, social presence positively influenced learner engagement, with instructor credibility fully mediating this relationship. According to a two-way MANOVA analysis used to examine the effects of humanness and nativeness on credibility and engagement, no evidence was found to support a difference between AI instructors and their human counterparts when observing learners’ perceptions and engagement, regardless of whether the instructors were NESTs or NNESTs. The results show that AI instructors can be a viable alternative in language classes.