Adoption of Artificial Intelligence in Library Operations and Service Delivery in Academic Libraries in Delta State, Nigeria

Artificial intelligence (AI) is reshaping how academic libraries acquire, organise, retrieve and deliver information, yet evidence on its uptake in Nigerian sub-national contexts remains thin. This study examined the adoption of AI in library operations and service delivery in academic libraries in Delta State, Nigeria. It determined the extent of AI adoption, tested the joint and relative influence of perceived usefulness, facilitating conditions and staff AI competence on adoption, and identified the challenges that constrain adoption. The study adopted a descriptive correlational survey design anchored on the Technology Acceptance Model and the Unified Theory of Acceptance and Use of Technology. A structured questionnaire (Cronbach's α = 0.80–0.91) was administered to professional and paraprofessional staff of university, polytechnic and college of education libraries, and 186 valid responses were analysed. Data were analysed with descriptive statistics, Shapiro– Wilk and Kolmogorov–Smirnov normality tests, Pearson correlation, collinearity diagnostics, multiple regression, Breusch–Pagan and White tests, the Durbin–Watson statistic, Ramsey's RESET, Cook's distance and 95% confidence intervals. The overall extent of AI adoption was low (M = 2.29, SD = 0.62, against a criterion mean of 2.50). Similaritychecking software, generative AI tools and AI-assisted discovery were the only applications adopted to a high extent; predictive collection analytics, robotics and chatbots trailed. The three predictors jointly explained 44.1% of the variance in adoption (R² = 0.441, adjusted R² = 0.432; F(3, 182) = 47.84, p < .001). Perceived usefulness (β = 0.341), facilitating conditions (β = 0.307) and staff AI competence (β = 0.270) each contributed significantly (p < .001). All regression assumptions were satisfied. Inadequate funding, the high cost of AI software and erratic power supply were the leading challenges. The study recommends dedicated AI budget lines, investment in power and bandwidth, structured AI-literacy training and institutional AI policies