Experiences of librarians with AI-driven recommendation systems in digital libraries: A phenomenological study

Abstract

This phenomenological study explores Nigerian academic librarians’ lived experiences with AI-driven recommendation systems in digital libraries. Using Colaizzi’s seven-step framework, semi-structured interviews were conducted with 12 librarians across Nigeria’s six geopolitical zones. Findings reveal that while AI systems enhanced workflow efficiency in cataloging and user services, librarians encountered critical challenges: algorithmic opacity hindered their ability to explain recommendations, cultural bias privileged Western over local content, and data privacy concerns persisted. Participants described a professional identity shift from content curators to system facilitators, accompanied by significant gaps in AI literacy and inadequate training support. The study underscores three imperatives for ethical AI adoption: (1) developing explainable, culturally responsive systems; (2) implementing continuous, localized training; and (3) involving librarians in system design and evaluation. Although limited by its small, Nigeria-specific sample, this research provides frontline insights into the tensions between AI efficiency gains and human-centered values. Findings call for coordinated action by developers, administrators, and policymakers to ensure transparent, equitable AI integration in Global South library contexts.