Despite the increased adoption of artificial intelligence (AI) chatbots in second language acquisition contexts, empirical examinations of learner perceptions regarding virtual AI-driven audiences remain sparse. This study utilizes mixed methodologies to analyze undergraduate students' experiences interacting with AI chatbot-driven virtual audiences during public speaking tasks in a virtual reality (VR) environment at a Japanese public university. Data collection comprised open-ended surveys juxtaposing student AI chatbot interactions with traditional peer audiences, triangulated with self-reported measures of digital literacy. Preliminary findings reveal pronounced variation correlated with student digital literacy competencies, which resulted in reduced public speaking anxiety, enhanced self-efficacy, and favorable appraisals of chatbot-mediated VR interactions, or increased anxiety and diminished comfort, largely attributing these reactions to perceived emotional responsiveness deficiencies in AI chatbots. The study underscores the critical necessity for differentiated instructional interventions to mitigate digital literacy disparities and optimize AI integration efficacy in immersive VR language learning environments. Recommendations include targeted digital competency training and iterative pedagogical scaffolding. Further research should systematically explore intervention strategies for effectively bolstering learner digital literacies, thereby facilitating meaningful engagement with emerging AI technologies in language instruction contexts.