Capabilities shaping institutional adoption: modeling generative AI adoption in higher education

Purpose Generative artificial intelligence (GenAI) is rapidly transforming higher education, yet its successful institutional adoption extends beyond technological capability to require effective knowledge governance, organizational readiness and coordinated decision-making. This study aims to structurally model the CSFs for GenAI adoption in education using an integrated interpretive structural modeling (ISM)–DEMATEL approach. The authors frame GenAI adoption explicitly as a knowledge management problem, analyzing how governance, infrastructure and privacy conditions enable or inhibit institutional capabilities for creating, sharing and governing GenAI-empowered knowledge. Design/methodology/approach The authors identified ten CSFs through a structured literature review and classified them as enablers, capabilities or outcomes across governance, infrastructure, data privacy, analytics, user experience and academic integrity. To analyze the causal and hierarchical relationships among CSFs, an expert validation was conducted through DEMATEL analysis by 20 experts, followed by ISM modeling, which involved 16 experts. Findings DEMATEL identifies technical infrastructure and scalability (CSF-V), data privacy and security (CSF-VII) and integration with LMS/SSO/enterprise systems (CSF-IX) as prominent causal factors. ISM further positions CSF-V at the deepest foundational level, followed by AI governance, policy and ethics (CSF-III), while assessment integrity and academic honesty (CSF-X) occupies the terminal dependent level. The findings identify assessment integrity (CSF-X), prompt literacy (CSF-IV) and user experience (CSF-VI) as dependent factors, confirming that they emerge from structural readiness rather than acting as independent adoption determinants. Research limitations/implications Results are restricted to expert judgment and defined CSFs, which can reduce generalizability across diverse educational settings. Future research must involve larger samples, validate empirically and use longitudinal designs to study changing relations while also testing the knowledge governance mechanisms revealed in this research. Practical implications The model enables phased prioritization of CSFs for optimal implementation of GenAI. Foundational readiness requires infrastructure, system integration and data governance, followed by governance policies and faculty capacity development, before addressing user experience and academic integrity as dependent outcomes. Originality/value The research provides a systems approach to GenAI adoption in education by presenting an integrated ISM–DEMATEL framework based on TOE and DOI, grounded in knowledge governance theory. It aids authors’ understanding of how technological, organizational and knowledge governance dimensions must be sequentially aligned to enable better implementation of GenAI.