Learner Perceptions of Artificial Intelligence-Generated Pedagogical Agents in Language Learning Videos: Embodiment Effects on Technology Acceptance

Abstract

Artificial intelligence generates vibrant characters, encompassing teachers, peer students, and advisors within diverse educational media. However, the impact of the perceived embodiment of such characters in language learning videos on students’ technology acceptance and adoption is unclear. Integrating structural equation modeling into thematic analysis, this study analyzes 1042 valid responses from higher education students to bridge this research gap. Our study reveals that four subdimensions of embodiment (human-likeness, credibility, learning facilitation, and engagement) significantly and positively predict higher-education students’ perceived ease of use and usefulness of artificial intelligence-generated virtual teachers in language learning videos. Notably, an exception arises, as human-likeness does not significantly predict students’ perceived ease of use in our research context. Students’ perceived systemic interactivity and impact on the learning process emerge as pivotal mediators. The qualitative thematic analysis identifies students’ concerns about classroom administration, developmental support, technical issues, deprived interpersonal collaboration, and liberal attainment cultivation with the virtual teacher presence. This study can illuminate artificial intelligence technology designs and applications in education.