Artificial Intelligence as a strategic driver for the transformation of library ecosystems from digital services to intelligent knowledge infrastructures

Abstract

The study examines artificial intelligence as a strategic driver of the transformation of library ecosystems against the backdrop of accelerating data growth and increasingly complex information demands, which makes the transition from digital services to intelligent knowledge infrastructures an urgent priority. The aim is to develop a quantitatively grounded model for AI implementation that enhances search relevance, personalizes interaction, and supports knowledge management. A systems approach with comparative analysis and modeling was applied; the empirical base includes 95 leading libraries from 18 countries over the period from 2019 to 2023. Strategies, reports, and software stacks were studied, and a survey of 250 specialists was conducted. For analysis, R and Python were used with LDA topic modeling and Word2Vec vector representations; Protégé and Neo4j were employed for ontologies and knowledge graphs. The Library Intellectualization Index (LII) is proposed as a weighted composition of routine process automation, cognitive service implementation, and analytical infrastructure maturity. The results show high completion of basic digitalization with an average value of 89.5 percent, low adoption of advanced AI services at 35.1 percent, and a scarcity of full-fledged intelligent infrastructures at 10.3 percent; research university libraries lead, while public libraries lag significantly. Semantic search based on BERT-type models demonstrates 85.7 percent precision and 82.3 percent recall with an average time of 55.9 seconds, statistically outperforming alternatives (p 0.001). Personalization increases engagement indices up to 0.79 and retention to 78.2 percent; session duration growth reaches 58.1 percent, with a nonlinear dependence. The correlation between basic automation and AI implementation is moderate (r=0.58); the regression model explains up to 72 percent of engagement variance, with key predictors linked to advanced service maturity and search accuracy. Risks of new technological inequality, the need for new competencies and ethical principles are discussed, and a roadmap for phased intellectualization and directions for scalable open platforms is proposed. The practical value lies in recommendations for forming employee competency profiles, implementing open metadata exchange standards, developing interoperable APIs and knowledge graphs, setting transparent quality metrics, managing privacy and bias risks, as well as planning funding and pilot stages. Prospects include open model repositories and collaborative consortia for the library sector.