Prioritizing the core dimensions of teacher AI literacy for AI-empowered teaching capability in higher education: a Delphi study

Generative artificial intelligence has permeated higher education, highlighting an urgent need to clarify core dimensions of teachers’ AI literacy for AI-enabled instruction. Current AI literacy frameworks for educators are overly general and fail to rank critical literacy dimensions suited to university teaching. Unlike basic AI operation, higher education demands teachers embed AI into disciplinary instruction while retaining independent pedagogical judgment, academic norms and human supervision. To fill this research gap, this study adopts a two-round Delphi method involving 22 experts to extract and prioritize core AI literacy dimensions for college teachers. Statistical results identify eight dimensions with moderate expert consensus (Kendall’s W = .35, p < .001). Pedagogical Integration & Teaching Innovation, Human–AI Collaboration & Boundary Awareness, and Contextualized Disciplinary Adaptation rank top three. Instead of adding another general competency catalogue, this research constructs a prioritized literacy framework for higher education, offering solid references for faculty training, institutional policy support and relevant measurement tool development.