Developing pre-service English teachers’ critical and ethical AI literacy in writing with generative AI

Abstract

This mixed-methods study explored the effects of AI literacy (AIL) training on pre-service English teachers’ writing proficiency and Al literacy development. Employing a quasi-experimental design (n = 60), the study compared an AIL training group (G1), a just-writing group (G2), and a control group (G3). All groups completed surveys measuring critical and ethical AI literacy, while G1 and G2 (n = 43) also completed writing tests. Qualitative data were collected from G1 through reflection journals and follow-up interviews (n = 7). A one-way analysis of covariance (ANCOVA), using pretest scores as a covariate to control for cohort effects, revealed that the AIL group significantly outperformed the just-writing group in post-intervention writing proficiency ( F (1, 40) = 13.29, p < .001). While self-reported surveys showed no significant shifts in critical or ethical AI competence ( p > .05), qualitative data substantiated the behavioral manifestation of AI competence. Students in the AIL group demonstrated critical competence through behaviors such as cross-checking AI-generated sources, and limiting AI use to brainstorming after experiencing hallucinations. Based on these findings, the study proposes two conceptual frameworks. The TPEC (Technological, Pedagogical, Ethical, and Critical) framework is presented as a theoretical expansion of existing AI integration models for teacher education. The TDCE model (comprising Technological, Dialogical, Critical, and Ethical dimensions) outlines a developmental trajectory of AI literacy from awareness to competence across the four dimensions. These models suggest that structured AIL training supports the responsible integration of generative AI into writing instruction, even when changes in attitudes and competencies are not immediately captured by self-reported questionnaires.