From Metrics to Meaning: A Three-Layer DCI Framework of Generative AI Adoption and Intellectual Governance

Generative AI is now embedded in higher education, yet research often treats adoption as a static metric. This study examines how students actively appropriate AI in a post-adoption landscape, reconceptualising traditional adopter categories as dynamic identity resources. Using a mixed-methods design, we analysed 188 reflective essays (framed as curated identity performances) from postgraduate Information Systems students via qualitative content analysis and K-medoids clustering. We propose the generative DCI Framework, bridging macro-level social diffusion with micro-level AI competence and learner identity to explain GenAI adoption as intellectual governance. Findings reveal distinct cross-layer mechanisms: while most identify as Early Adopters or Early Majority, they differ sharply in how they legitimise use through verification, boundary-setting, and identity protection against career-readiness pressures. The study contributes empirically informed typologies to guide stakeholder-specific AI literacy initiatives and process-oriented assessments.