Development and Preliminary Validation of a Self-Report Scale on Generative AI Literacy in Learning Among High School Students

Le résumé fourni par la source

The use of generative AI (GenAI) in learning creates a need for assessment instruments that are appropriate to learners’ developmental characteristics, learning purposes, and educational contexts. Existing scales, however, have largely been developed for higher education or non-Vietnamese contexts, whereas high school students differ in learning autonomy, academic integrity experience, and technology-use conditions. This study develops and preliminarily validates a self-report scale assessing high school students’ GenAI literacy in learning. The scale was developed from existing AI/GenAI literacy frameworks, revised Bloom’s taxonomy, and Vietnam’s digital competence framework for learners. Survey data from 1,003 students at eight high schools in Ho Chi Minh City were analyzed using Cronbach’s alpha and Exploratory Factor Analysis. The findings provide initial evidence of reliability and a multidimensional structure consisting of 23 observed items across five dimensions: Knowledge & Understanding, Use & Application, Evaluation, Create, and Ethics. The psychometric evaluation yielded a KMO value of 0.896, a significant Bartlett’s test (p < 0.05), and a total variance explained of 59.07%, while internal consistency was confirmed with Cronbach’s alpha coefficients ranging from 0.691 to 0.851 across the factors. The study proposes a contextually grounded instrument that may help schools identify students’ self-perceived GenAI literacy. Beyond its methodological contributions, this scale serves as a pivotal diagnostic tool for educational stakeholders to tailor AI integration strategies and for policymakers to refine digital literacy standards within the general education sector, while emphasizing the need for further validation through CFA, convergent and discriminant validity testing, and measurement invariance before broader use.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.