Designing Fairness: Human-Centered AI Use in Digital Libraries for Ethical Knowledge Access

As artificial intelligence (AI) increasingly shapes digital library services, ethical, transparent, and human-centered implementation has become a critical concern. This qualitative case study explores how Human-Centered AI principles—including fairness, transparency, user agency, and accountability—are understood and enacted in AI-enabled digital libraries in Bangladesh. Data were collected through semi-structured interviews with 20 librarians, systems developers, faculty members, and students from five public and private universities. Thematic analysis found that AI-supported discovery, automated cataloguing and metadata generation, recommender systems, and chatbot-based reference services are increasingly integrated into digital library environments. However, ethical implementation remains uneven due to limited governance frameworks, dependence on vendor-supported or open-source platforms, resource constraints, and differing stakeholder priorities. Librarians emphasized governance, developers focused on technical feasibility and explainability, faculty highlighted scholarly trust and information quality, while students prioritized usability and transparency. The findings demonstrate that ethical AI implementation is a socio-technical process shaped by stakeholder roles, organizational governance, technological capability, and institutional context. The study contributes context-sensitive evidence to Human-Centered AI and Library and Information Science scholarship and highlights the need for institution-specific governance, ethical capacity building, participatory design, and user awareness.