Rethinking the Structure of AI Literacy in Generative AI Educational Systems: Descriptive Patterns of Expert Convergence and Heterogeneity

Abstract

The integration of generative AI into education raises questions about how AI literacy should be organized, particularly as competencies may operate across different functions during human–AI interaction. This study examined the content and structural organization of AI literacy through a focused literature synthesis and a two-round Delphi study with 11 experts. An initial framework comprising three domains—AI Understanding, AI Application, and AI Integration—and nine competencies was developed and iteratively evaluated. Experts generally evaluated the competency content favorably, while their judgments showed greater heterogeneity regarding some aspects of structural representation. In Round 2, experts supported non-exclusive domain relationships (M = 4.45, SD = 0.52) and primary and secondary domain associations (M = 4.73, SD = 0.47), whereas the proposed representation of competency placement and cross-domain applicability showed greater dispersion (M = 3.73, SD = 1.19, IQR = 1.50). This descriptive pattern informed a theoretical interpretation of the three domains as functional rather than mutually exclusive categories. The resulting framework provides a basis for considering how AI literacy competencies may be enacted during human–AI interaction and related to curriculum, pedagogy, assessment, teacher roles, and AI technologies in generative AI-supported educational systems.