Improving the quality of AI-supported K–12 teaching: the effects of teachers’ AI-TPACK and self-efficacy

Artificial Intelligence (AI)-supported teaching is gaining popularity in K–12 schools. However, there have been few empirical investigations on how to improve its quality. To address this gap, this study examined two antecedents of K–12 teachers’ self-reported instructional quality, namely AI-technological pedagogical and content knowledge (AI-TPACK) and self-efficacy in using AI to empower learners (SEEL). Specifically, SEEL is conceptualized in two subdimensions: self-efficacy in using AI to actively engage learners and self-efficacy in using AI to promote differentiation. Employing a proportionate stratified random sampling procedure, 931 teachers from nine K–12 schools in different areas were surveyed. Structural equation modeling showed that K–12 teachers’ AI-TPACK and self-efficacy in using AI to actively engage learners significantly impacted instructional quality, while the impact of self-efficacy in using AI to promote differentiation was not significant. The mediation effects of AI-TPACK on instructional quality via self-efficacy in using AI to actively engage learners were also confirmed. This study theoretically revealed two significant predictors of the instructional quality of AI-supported teaching and has important practical implications on how to improve the quality of AI-supported teaching.