From digital libraries to future libraries: Reframing academic libraries as knowledge infrastructures in the age of artificial intelligence

Abstract Generative artificial intelligence (AI) is reshaping the production, organization, dissemination, and use of knowledge. In this context, the traditional positioning of academic libraries as knowledge repositories is increasingly challenged by commercial AI platforms, while growing disintermediation raises questions about their role in contemporary knowledge ecosystems. Drawing on Infrastructure Studies and socio-technical systems theory, this study explores the conceptual transition of libraries from knowledge repositories to knowledge infrastructures. To support this argument, it proposes a “Four Pillars, Five Wings” framework, comprising four foundational capacities (AI model base, cross-media knowledge resources, full-chain knowledge services, and digital–physical integrated spaces) and five emerging capabilities (knowledge discovery, intelligent learning, collaborative creation, cultural activation, and research augmentation). Using Zhejiang University’s Future Library initiative as an empirical case, the study examines implementation pathways and discusses key challenges encountered during the early stages of development. The findings contribute to ongoing discussions on the transformation of academic libraries from resource-centered institutions into knowledge infrastructures in the age of artificial intelligence.