AI for Readers' Advisory

Readers’ advisory connects readers with books that match interests, moods, reading preferences, or curiosity. AI can generate read-alike lists, describe themes, or ask follow-up questions that help refine a search. Recommendations rely on available metadata and patterns, which may overrepresent popular titles, misread genre, or fail to capture why a reader liked a particular work. A generative system can invent books, authors, plot details, or publication facts. Readers should verify titles and availability in a trusted catalog or publisher source before relying on the list. Libraries should make it clear whether a recommendation comes from an AI tool, preserve non-personalized ways to browse, and offer human assistance. Reading histories and preference profiles can be sensitive, so systems should collect only needed information and explain retention and personalization. The American Library Association’s AI guidance recommends transparency, privacy protections, and preserving predictable discovery paths alongside AI-enhanced recommendations. Evaluation should consider whether recommendations are real, relevant, diverse, and responsive to corrections, not simply whether users click. Recommendation systems can shape which authors and viewpoints become visible. Librarians can help interpret ambiguous requests and bring knowledge of collections and communities. AI may provide a starting point for exploration, but a reader should remain able to choose, correct, and discover books beyond the system’s ranking. Libraries should avoid inferring sensitive traits from sparse reading activity.