Speaker
Description
This study examined the factor structure and predictive value of four scales measuring AI literacy, writing performance with generative AI (GAI), GAI-driven well-being, and academic writing ability among 51 Vietnamese EFL learners at a private language centre. An 18-item adapted Likert questionnaire was analysed using Exploratory Factor Analysis, Cronbach's alpha, correlation, and multiple regression. Four factors emerged, accounting for 63.8% of variance: Academic Writing items loaded cleanly, but Writing Performance and Well-Being merged into one factor, and AI Literacy items were dispersed, with marginal reliability. AI literacy was the only significant unique predictor of academic writing ability, while performance and well-being measures showed no independent contribution once it was accounted for. These findings suggest that explicit instruction in AI literacy, covering prompting, evaluation, and integration skills, may support learner agency in AI-assisted writing more effectively than generic GAI exposure, and that current well-being and performance measures need refinement before they can meaningfully inform classroom-level assessment of AI-supported academic writing.
