AI hallucinations are causing confusion among library patrons, leading to requests for non-existent books and diminishing trust in human librarians' expertise, as patrons increasingly rely on AI outputs.

Librarians, such as reference librarian Eddie Kristan, are increasingly asked to find non-existent books that are hallucinated by AI, a problem worsened by AI-generated fake book lists circulated in major newspapers earlier this year.

Patrons tend to trust AI-generated information more than human librarians, which reduces critical thinking, curiosity, and raises mental health concerns, according to surveys and librarian reports.

Vendors are pushing AI products into libraries without sufficient testing, resulting in issues like poor accuracy and the erosion of traditional research skills, exemplified by tools like AI Insights and retrieval-augmented generation.

While some librarians remain cautiously optimistic about AI's potential to enhance information retrieval, current implementations are often clunky, error-prone, and disorienting for both staff and users.

Kristan emphasizes the need to advocate for the profession by highlighting AI's negative impacts on the community and seeking equitable solutions to preserve information literacy and trust.

Library staff are requesting vendors to remove AI-generated titles from digital collections and are wary of AI tools that undermine basic research skills, such as precise searching.

Kristan employs methods like checking library catalogs such as WorldCat and tracking sources like Kindle Direct Publishing to identify AI-generated or non-existent titles.

AI tools are infiltrating library systems and content, with vendors rushing to integrate large language models into search functions and summaries, often misleading users about their capabilities.

Many AI-driven library tools, including semantic search and AI summaries, are flawed; they often claim to understand user intent or improve search accuracy but frequently produce inaccurate or overly broad results.