Exploring Vietnamese EFL learners’ generative AI usage patterns, self-efficacy and self-directed learning: Preliminary findings

Abstract

While AI tools are increasingly integrated into language education to support learning autonomy and proficiency, limited research has explored how Vietnamese EFL learners perceive and engage with AI in relation to their self-efficacy and self-directed learning. This study surveyed 100 Vietnamese undergraduate English majors to examine patterns of AI engagement and their associations with self-efficacy and self-directed learning behaviors. Data were collected via an online questionnaire that measured students’ AI knowledge, usage frequency, and perceived understanding of AI-generated content. Descriptive and inferential analyses showed that 89.89% of participants reported at least basic AI knowledge, though only 8.08% identified as having advanced proficiency. ChatGPT (39.81%) and Grammarly (22.69%) emerged as the most frequently used tools, mainly for translation (21.84%) and writing assistance (18.35%). Despite frequent use, students’ understanding of AI remained surface-level, with only 1% having studied AI models and nearly 70% lacking deeper comprehension of how AI systems function. A statistically significant association was found between GPA and AI use frequency (F = 2.65, p = .038), while no significant differences were observed based on gender or year of study. Notably, a strong correlation was observed between self-efficacy in using AI and self-directed learning with AI (r = 0.71, p < .001), suggesting that students who feel more confident in their ability to use AI tools are also more likely to engage in autonomous, AI-supported learning. These findings underscore the value of incorporating AI literacy and critical engagement strategies in EFL instruction to better support learners' autonomous practices.