Artificial intelligence in libraries: a systematic review of upskilling, challenges, and future research directions

Introduction. Libraries, as trusted institutions, are positioned to advance AI literacy for both staff and patrons. This study investigates how AI literacy has been conceptualised and implemented in libraries and examines the challenges these institutions face in AI upskilling and integrating AI into their services. Method. A systematic literature review using PRISMA guidelines to identify peer-reviewed articles and practitioner reports on AI upskilling in libraries. Analysis. Fourteen articles published between 2021 and 2025 were examined, spanning academic, public, and school libraries. Both thematic and comparative analysis were used to synthesise findings. Results. Findings reveal significant variation in definitions of AI literacy, sectoral differences in focus (staff training in academic libraries, patron programs in public libraries, early digital citizenship in schools), and recurring challenges including cost, vendor dependence, staff anxieties, and lack of outcome evaluation. Geographic patterns show Asia leading in empirical studies, while Western literature is largely conceptual. Conclusion. To remain relevant, libraries must move beyond fragmented, vendor-driven initiatives toward clearer definitions, equity-centered approaches, and scalable, evidence-based models that empower both staff and patrons in an AI-driven era.