Abstract
While multimodal generative artificial intelligence (GAI) is increasingly integrated into educational contexts, empirical evidence remains limited regarding how intentionally designed GAI-supported learning environments can be associated with multimodal GAI literacy and intercultural communicative competence (ICC). This convergent mixed-methods study investigated a multi-stage intercultural learning design involving students from China, Northern Cyprus, and Kazakhstan (N = 71). Four pedagogical agents guided learners through multimodal production tasks, including digital human self-introduction videos, culturally contextualized image generation, and multimodal cultural videos development. Structured reflections were facilitated following all activities. Paired-samples t tests revealed significant pre-post gains across all dimensions of multimodal GAI literacy and ICC. Performance-based ratings of learners’ multimodal artefacts further demonstrated progressively higher design quality across tasks, suggesting improved artifact performance across production contexts. Qualitative findings indicated that learners moved beyond operational tool use toward strategic multimodal design, critical evaluation of AI outputs, calibrated trust, and communicative mediation, alongside contextualized cultural understanding and strengthened intercultural reasoning. The findings suggest that structured multimodal GAI integration may be associated with technological, creative, and intercultural development. By triangulating survey, artifact, and qualitative evidence, this study advances a process-oriented conceptualization of multimodal GAI literacy and provides design-informed insights for integrating GAI into intercultural learning.