Exploring the prediction of English learning self-efficacy by behavioural intention to use AI, dangerous beliefs in AI and AI skills

Artificial intelligence (AI) is increasingly shaping students’ English as a Foreign Language (EFL) learning. This study applied the Unified Theory of Acceptance and Use of Technology (UTAUT) to explore AI attitudes for fostering English learning self-efficacy. A survey of 612 senior high school students confirmed the reliability and validity of eight constructs: English learning self-efficacy (SE), dangerous beliefs in AI, AI skills, behavioural intention to use AI, effort expectancy, performance expectancy, social influence and facilitating conditions. Results indicated that behavioural intention, dangerous beliefs and AI skills significantly predicted self-efficacy, confirming the direct prediction of students’ attitudes towards AI on EFL learning self-efficacy. Although dangerous beliefs did not significantly predict behavioural intention or AI skills, AI skills predicted behavioural intention and were predicted by social influence and facilitating conditions. Social influence emerged as a key factor, predicting dangerous beliefs, AI skills, behavioural intention and broader AI perceptions. These findings highlight the essential role of social context in AI adoption and suggest rethinking the functions of dangerous beliefs in AI and AI skills in AI-supported environments. These findings offer theoretical and practical insights for researchers, teachers and policymakers to enhance AI integration in EFL by addressing students’ learning self-efficacy and AI perceptions.