Digital application literacy as a mediator of GAI-TPACK among non-STEM teachers after year-long professional development

Generative artificial intelligence (GAI) can produce and revise instructional content through probabilistic, conversational interaction, creating demands that extend beyond routine use of digital tools. This study examined how digital technology knowledge and skills (DL-S) and digital application literacy (DL-A) relate to GAI-integrated technological pedagogical content knowledge (GAI-TPACK) among 224 Chinese primary and secondary teachers in general-subject settings. Participants completed a year-long professional development program and a post-intervention survey. Structural equation modeling showed positive paths from DL-S to DL-A (β = 0.300) and GAI-TPACK (β = 0.317), and from DL-A to GAI-TPACK (β = 0.453). The indirect effect of DL-S through DL-A was 0.137 (95% bootstrap CI [0.069, 0.219]), consistent with partial mediation. The larger point estimate for DL-A was not tested directly against the DL-S coefficient and is therefore interpreted descriptively. These findings position DL-A as a plausible practice-oriented mechanism linking foundational digital capability with subject-specific GAI integration. Because competence was self-reported at one post-intervention time point, the model does not establish causal development. The study offers a bounded operational account of GAI-TPACK for non-STEM professional development and identifies priorities for multi-wave and performance-based research.