The generative AI turn in libraries: Mapping emerging research, AI literacy, and service innovation

Artificial intelligence was integrated into library and information science before the emergence of ChatGPT. Earlier research on AI in libraries emphasized automation, smart libraries, metadata management, information retrieval, recommender systems, chatbot services, AI adoption, and professional readiness. In the literature, AI was primarily conceptualized as a technical layer supporting library systems and service operations. The advent of generative AI has altered this perspective by positioning large language models closer to search, reference interactions, research support, information literacy instruction, chatbot services, and metadata creation. This study investigates whether generative AI extends earlier AI-enabled library automation or indicates a new research frontier in AI literacy, service innovation, and responsible use. A comparative, two-layer science-mapping design was employed, using Scopus as the data source. Dataset A encompasses AI-in-libraries literature from 2000 to 2026, while Dataset B includes post-ChatGPT generative AI literature in libraries from 2022 to 2026. Both datasets were independently screened and analyzed before cross-layer comparison. Biblioshiny and HistCite were used for bibliometric mapping. The findings indicate both continuity and repositioning. Dataset B remains connected to earlier AI-in-libraries research, particularly information retrieval, natural language processing, metadata, academic libraries, and library services. However, generative AI-related themes are more closely associated with user-facing activities. This shift suggests a movement from background infrastructure toward user-oriented knowledge mediation.