Curating Trust and Connections: Transforming the Information Ecosystem with Instruction, Grounded Curation, and Open Pedagogy

Abstract

In an era of context collapse and a disrupted information landscape, where the origins and authority of information are increasingly obscured, this presentation demonstrates how Cal Poly Humboldt Library is advancing AI literacy and skills, while also developing innovative trusted digital environments. By leveraging the expertise of library professionals in curation and information literacy, we can transform AI from a source of misinformation into a powerful tool for discovery and learning.

This session presents a multifaceted case study from Cal Poly Humboldt, highlighting a spectrum of AI integration, from AI Worldcafés, to AI workshops, to the creation of custom, source-grounded AI applications within an open pedagogy framework. The instructional approach ranges from Slice of AI & Pie, to AI SkillShops, designed to help students, staff, and faculty move from using general AI chatbots to utilizing grounded systems. Unlike general models that rely on broad internet datasets, the instruction focuses on finding and uploading trusted sources in AI tools. These AI Skillshops address a need to share techniques to build trust and AI literacy using grounded tools informed by user-uploaded trusted sources that improve AI citing directly to high quality works, enhancing critical thinking, and reducing opportunities for hallucinations and unsupported claims.

The presentation further details the development of Ask Alex Humboldt, a collaborative research project that utilizes a Retrieval-Augmented Generation (RAG) pipeline. This application leverages the library’s Digital Commons institutional repository as knowledge source to a vector database, connecting academic repositories to large language models. Our goal is to ground AI responses in peer-reviewed scholarship and specialized open science repositories, our Library seeks to add meaningful connections of academic research with community science and observations.

Crucially, these projects employ an open pedagogy approach, involving students as active co-creators in the design and testing of agentic AI. By working under the constraints of open-source and proprietary models, students learn the mechanics and realities of AI; such as document parsing with Docling and parallel processing with Ray clusters, while grappling with the ethical implications of algorithmic bias and data privacy. This hands-on collaboration empowers students to move from passive consumers of AI to informed architects of trusted information systems.

Ultimately, this session demonstrates that the role of libraries in the age of AI is not just to teach literacy, but to lead the development of scholarly-informed AI agents. By combining expert curation with agentic AI design, libraries can strengthen authority and transparency within the digital information environment, ensuring that the future of research remains both innovative and trustworthy. Keywords: AI literacy; grounded curation; retrieval-augmented generation; open pedagogy; information trust