Preservice Early Childhood Teachers’ AI Literacy and Self-Efficacy in a Structured “Vibe Coding” Design Task: An Exploratory Pre–Post Study

This exploratory study examined a structured Vibe Coding task in preservice early childhood teacher education and its association with self-reported AI literacy, self-efficacy, attitudes toward AI, and technology acceptance. Vibe Coding was used in a qualified sense: participants created educational applications through natural-language prompting, then reviewed and revised the outputs. The task was implemented in two undergraduate courses using a convergent mixed-methods, single-group pre–post design without a comparison group. Quantitative analysis included 32 matched cases, while qualitative analysis drew on 26 reflective logbooks. All 32 participants also completed a researcher-designed, non-validated self-assessment rubric, reported descriptively. Paired-samples t-tests showed significant pre–post increases across all six self-report scales after Holm–Bonferroni correction. Effect sizes were large for AI literacy, self-efficacy, attitudes, and perceived ease of use, and moderate for perceived usefulness and behavioural intention. These changes indicate perceptions rather than demonstrated competence and cannot be attributed causally to the task. Directed qualitative analysis identified three predefined categories: pedagogical empowerment, prompting and technical confidence, and professional identity reconfiguration. Participants described refining prompts, identifying errors, and justifying revisions against pedagogical criteria. The findings provide exploratory evidence of how preservice teachers learned to evaluate and improve AI-generated educational outputs through a structured design task.