Can Universities Forge Organizational Readiness for AI Adoption? Evidence from Meta-Analysis
TL;DR
The strongest effect is shown by individual determinants, which confirms the behavioural logic of technology adoption models in which intention is the closest proximal predictor of the actual introduction of AI in HE, and future studies should standardize readiness measurement and test contingent and mediating mechanisms in HE AI implementation.
Abstract
Nowadays, universities eagerly aim for artificial intelligence (AI), yet empirical evidence remains fragmented across different levels of analysis, AI system types, and implementation contexts, blurring cumulative conclusions. Empirical models remain limited, lacking big picture that could give complex understanding of how implementation processes really work out in modern universities. This study aims to identify determinants of organizational readiness for AI adoption in higher education via meta-analysis and consolidate existing predictors into semantic blocks. A systematic search of literature yielded 28 empirical HE studies using PLS-SEM. Following PRISMA, predictors were coded into three levels and random-effects meta-analysis was conducted in R; pooled effects with confidence intervals were estimated and heterogeneity assessed using Q, I², and τ². Twelve out of fourteen observed predictors showed statistically significant positive effects on AI adoption. The strongest effect is shown by individual determinants, which confirms the behavioural logic of technology adoption models in which intention (ES = 0.907; p <0.001) is the closest proximal predictor of the actual introduction of AI in HE. Cognitive predictors such as perceived usefulness and perceived ease of use also show comparable sustained effects, which confirms the importance of evaluating individual, cognitive-behavioural characteristics of personnel when implementing AI in HE processes The importance of capability-oriented factors was highlighted with AI literacy and self-efficacy, showing the importance of professional development programs for teachers to adopt AI successfully. Institutional determinants, such as social influence and competitive pressure have also shown significant results, indicating the importance of the external regulatory environment for universities. Organizational determinants also provided significant results. High heterogeneity for organizational readiness (I² = 0.84) indicated variation by context, AI solution type, and operationalization, supporting readiness as a dynamic capability. Given conceptual fragmentation and inconsistent modelling roles, future studies should standardize readiness measurement and test contingent and mediating mechanisms in HE AI implementation.