Rethinking reference in the AI era  

The nature of reference librarianship becomes transformational as knowledge practices change, trust is renegotiated, and user questions shift, all thanks to AI.


A series of lightning talks by global librarians working in several different types of libraries highlighted how reference librarians are responding to emerging AI technologies. The constant is that reference librarians still strive to connect users to appropriate resources. However, the nature of reference queries is changing, as users increasingly want to know how to use AI tools responsibly and librarians look at innovative ways to incorporate AI. Sources are also changing, with AI-generated content creating concern. Overall, the talks, given at IFLA’s World Library and Information Congress (2026.ifla.org) 2026, held in Busan, South Korea, had a positive spin on the transformation of libraries in the AI era.

In the public library sphere, explained Hsiang Ping Ma, Director General, National Library of Public Information, Taiwan, an AI-powered librarian-like interface was installed in a public library lobby. Named Xiaoshu, she is in a semi-enclosed booth, responds to voice input, and gives voice and text responses. First time users could compare AI with existing services, recognize different strengths, and position AI within the library ecosystem.

Crystal Yin, Suyu Dong, and Brianna Henshaw, Justice Institute of British Columbia, Canada (JIBC), looked at the evolving role of the reference librarian in AI literacy. To address a growing number of AI-related questions, their library introduced a three-part AI workshop focused on teaching the fundamentals of how AI works. They believe that reference services need to move beyond teaching the research process toward helping users understand and navigate the broader information landscape. Librarians heed to understand the technology before they can adequately teach it. Human-centered reference support is still needed. AI literacy should be individualized across all types of reference activities.

AI tool assessment, according to Haoyong Lan, STEM faculty librarian at Carnegie Mellon University (CMU), United States, is an emerging reference practice. It’s an extension of core reference skills, not a new domain. Evaluating AI tools follows the same logic as evaluating any source: credibility, relevance, and fit. Librarians need a working knowledge of retrieval, generation, output quality, accessibility, and sustainability, which provides a structured framework and gives librarians a shared vocabulary and consistent approach to assessment.

Martian Jinio and Nikko Angelo Lastimosa, University of the Philippines Diliman, asked, "If AI now answers first, what happens to reference work?" Before users consult a librarian or open a database, many have already asked ChatGPT, Gemini, or another AI tool. Sometimes the answers are useful. Sometimes they are wrong. And sometimes they are wrong in ways that look completely believable. Jinio and Lastimosa identified four themes for reference librarian work: verification and repair; workflow adaptation; trust, authority, and mediation; and selective and context-dependent AI use. These themes are not only about technology; they are about labor, trust, and how librarians negotiate information in environments where information is increasingly easy to generate. The practical implications are the strengthening of AI literacy, the development of verification workflows, the creation of guidance on hallucinated citations, and the supporting of librarian mediation. Reference work is changing, not disappearing.

Yu-gyung An, Research Support Librarian, Seoul National University, South Korea, drew upon 19 years of research support data and concluded that the evolving role of librarians lies in the ability to choose rather than simply providing an answer. Mediated search requests have dropped dramatically. Judgment-support services, such as methodology consultation and journal selection, that did not exist in the early data, have grown as researcher needs shifted. Document delivery has stayed at a constant level. The librarian is no longer the person who finds information, but is now the person who helps researchers make decisions. AI raises a basic question: if technology can search, analyze, and even write, then what are libraries for? The mission of reference librarians—connecting people to information—has not changed. Only the environment around it has.

Following the talks, attendees discussed how reference services have changed, new competencies, alterations in daily practice, and mitigating the fear factor in their libraries.