From Resistance to Complementarity: A Comparative Analysis of Organizational Readiness for AI in Italian Public Universities

Purpose. The rapid integration of Artificial Intelligence (AI) into Higher Education Institutions (HEIs) challenges traditional administrative workflows, demanding a shift from individual adoption to organizational readiness. While previous literature has largely relied on single-case studies, this paper aims to bridge the gap by conducting a comparative analysis between two Italian public universities: the University of Urbino Carlo Bo and the University of Ferrara. The study investigates how organizational context, job hierarchy, and digital competence influence the perception of Human-AI complementarity. Methodology. The research employs a quantitative approach based on a survey administered to a total sample of N = 532 administrative and technical staff members (NUrbino = 231; NFerrara = 301). The methodology utilizes Independent Samples t-tests to compare institutional contexts and Conditional Mean Analysis to explore hierarchical variances. A multi-dimensional “AI Literacy Index” was constructed to test correlations with perceived risks, specifically “Job Replacement” and “Cognitive Obsolescence”. Finally, a cluster analysis segmented the workforce into psychographic profiles. Findings. Results highlight a significant “context effect”: the University of Ferrara exhibits higher AI Literacy (p = 0.002), which correlates with significantly lower anxiety regarding cognitive decline compared to Urbino (p = 0.011). However, the “Fear of Job Replacement” shows no significant difference between the two institutions (p > 0.05), suggesting that public sector stability acts as a universal buffer against automation anxiety. A vertical “Digital Divide” emerges: High Qualification officials (EQ) demonstrate the highest competence and lowest fear (Mean = 2.25), whereas Library staff and “Collaborators” represent the most vulnerable clusters (Mean = 2.68), perceiving AI as a direct threat. Additionally, a persistent gender gap in technical literacy was identified. Correlations confirm that higher literacy reduces fear (r = –0.181, p < 0.001). Overall, by moving beyond single-case limitations, this study provides robust empirical evidence that one-size-fits-all training policies are ineffective. The cluster analysis identifies that 29% of the workforce are “Sceptics” (Low Competence / High Fear), requiring urgent technical reskilling, while only 27% are “Champions”. The paper offers university management a differentiated framework to foster a symbiotic Human-AI workforce, suggesting that training must be tailored by role function rather than just administrative level.