Unexpected AI in the Collections: A Collection Development Case Study on Navigating a New Academic Information Ecosystem

Abstract

Use of generative AI in scholarly published content continues to be experimental and library metadata standards are not being developed fast enough to deal with the variety of machine-generated content that is emerging. A new type of ebook was found in our library catalog, a human-mediated repackaging of AI summarized content, without prior awareness or selection by staff. In response to this example, librarians across the University of California Libraries held discussions around patron needs, vendor relations needs, and collections development needs to address considerations for new collections processes and workflows. Librarians noted they’re frustrated by this repackaging of existing materials to be resold back to libraries, which requires more work and effort for staff to mediate or circumvent. In conversations with this global academic publisher and our books marketplace vendor, we advocate for more notification and identification tools of machine-generated content to prepare for a future where permutations of machine-human content are likely to arise. Although AI-generated content is now mentioned in several public library collection policies, it hasn’t yet been adopted in many academic library guidelines. As AI literacy remains a main focus for library services, it is important to ensure that our library collections practices are aligned with this need, and to ensure authenticity of the information we offer. This case study shares ways in which our academic library plans to work with the machine-generated content in our collections as the scholarly publishing landscape evolves.

(presented on 15 August 2025 at "Curating in the Age of Generative AI: Global Perspectives on Collections, Ethics, Ownership, and Cultural Responsibility" session)