Domain-Differentiated Generative AI Competency Self-Efficacy Among University EFL Teachers: Patterns and Predictors

Self-efficacy, a person’s belief in their capability to carry out a given task, shapes whether teachers attempt, persist with, and transfer new practices. Yet teachers’ perceived capability to use generative AI (GenAI) is often treated as a single general readiness rather than examined across the distinct domains in which it is now used. Guided by Bandura’s self-efficacy theory, this study provides a domain-differentiated account of GenAI competency self-efficacy among university teachers of English as a Foreign Language (EFL) in the Saudi higher-education context. A sample of 232 EFL teachers at Saudi public universities completed a contextually adapted version of the Generative AI Competency Self-Efficacy Scale for Teachers and Researchers (GAICS-TR). Data were analysed using repeated-measures analysis of variance, correlational analyses, and hierarchical multiple regression. Overall self-efficacy was only slightly above the scale midpoint, and the overall score concealed an uneven pattern across domains. Teachers felt most capable in Basic GenAI Understanding and Teaching with GenAI, less so in Professional Engagement, and least confident in Research with GenAI, with the lowest confidence reported for data analysis. Prior AI training and frequency of GenAI use were associated with overall self-efficacy, whereas teaching experience and academic rank were not. The findings suggest that overall measures of GenAI readiness can mask lower-confidence areas within specific domains, especially in research-facing uses such as data analysis. Accordingly, supporting teachers’ development in using GenAI calls for targeted, domain-sensitive professional development that reaches teachers at all career stages rather than broad familiarisation with the tools.