Getting Beyond Verification: Reading Generative AI as an Unreliable Narrator
Most students have a vague sense that generative AI systems are unreliable, but they do not understand why. When we focus only on verification in AI literacy instruction we miss an opportunity to give them a fuller, more useful understanding of how the conditions of production influence the generative AI outputs that they encounter. Generative AI models are trained on a huge diversity of human-authored texts and media, yet their outputs are presented to the user through a seemingly singular, neutral and authoritative voice that obscures their epistemic instability. This workshop introduces an interpretive framework conceptualizing generative AI output as being produced by a polyphonic, unreliable narrator. This metaphor makes the sociotechnical processes that shape generative AI outputs more available to critical investigation. Participants will practice four pedagogical strategies (mapping the polyphony, provenance chasing, interrogating the narrator, and narrator/audience switching) that help students interrogate generative AI outputs and make more informed decisions about how they use them.
DAY: Wednesday, November 11, 2026
TIME: 12pm EST/ 5pm GMT/ 10:30pm IST
REGISTER: Click here (coming soon)
Featured Presenter
Stefani Gomez, PhD, MLIS is an Assistant Professor and User Experience Librarian at Kutztown University, where she develops and assesses introductory AI literacy for first-year seminar and composition courses. Her research and teaching explore the cultural and epistemological consequences of generative AI on the information ecosystem. She is currently working on developing pedagogical techniques that provide students with critical reading strategies for generative AI that remain useful even as these technologies quickly develop and change.