Large Language Model Literacy Framework for Librarians: Leading Information Stewardship in the Generative Artificial Intelligence Era

Abstract

Large language models (LLMs) fundamentally challenge traditional information literacy frameworks through unreliable information generation, bias perpetuation, and disrupted authority structures. As these systems become integral to research, education, and decision-making, librarians should lead information stewardship in the AI era by applying core professional expertise in information evaluation, collection development, and reference intermediation to LLM environments. This paper presents a comprehensive framework positioning librarians as uniquely qualified LLM literacy leaders through established LIS competencies. It treats LLMs as information resources requiring systematic evaluation, applies reference techniques to AI interactions, and maintains professional ethics concerning intellectual freedom and equitable access. Grounded in recent literature, the proposed framework defines core competencies, outlines implementation pathways, and presents assessment strategies that support librarians in guiding communities through AI-mediated information environments. It demonstrates how established LIS professional skills provide an essential foundation for democratic information stewardship, ensuring that LLMs serve community needs while preserving standards of critical thinking. This positions LLM literacy as a natural extension of the librarian’s professional identity, offering a forward-looking model for leadership in technologically mediated environments.