AI literacy profiles for human–AI collaboration in a GenAI-enabled learning environment: governance readiness and perceived course outcomes

Generative AI (GenAI) is increasingly embedded in university study as a dialogic partner, becoming part of students’ everyday study workflow. Yet learners in AI general education vary in what they report being able to do with GenAI and in how prepared they are to use it in policy-aligned ways. Using an anonymous end-of-semester survey of undergraduates (N = 617) in an AI general education course, we derived five self-reported AI literacy indicators (knowledge, practice, thinking, ethics, and basic digital skills) and applied latent profile analysis. An interpretable five-profile reference solution revealed a primarily low-to-high capability gradient. BCH three-step comparisons showed that profiles with higher self-reported capability reported stronger scores on a two-item governance readiness proxy (policy awareness and preference for compliant institutional tools) and stronger perceived course outcomes (perceived learning gains, satisfaction, and intention for further study). In contrast, broad GenAI usage scenarios were common across profiles, suggesting that the adoption-versus-sophistication distinction is a plausible interpretation rather than a directly observed process mechanism. We synthesize the findings into a capability–governance typology that supports profile-aware scaffolds, governance-embedded workflows, and process-oriented assessment for GenAI-enabled learning environments.