Wikimedia+Libraries International Convention 2025/Programme/information-literacy-ai-wikipedia
Rethinking Information Literacy in the Age of Generative AI: Can Wikipedia help?- ANN MATSUUCHI
Time: 12:00-12:15
Room: Library Lobby
Abstract:
Online research is increasingly shaped by commercial forces that prioritize SEO and AI harvesting, relying on a tracking-surveillance-attention model driven by data collection, user monitoring, and competition for attention. This model tailors content to users, often narrowing perspectives, while scrutinizing online behavior, potentially infringing on privacy. Our attention is persistently manipulated, diverting it from meaningful endeavors. With AI-guided filters and text generators now influencing content, it’s becoming harder to assess credibility and accuracy. Risks include AI-generated websites designed to manipulate search engine algorithms, unsupervised AI-authored articles that conflate similar topics or individuals, black-box filtering that obscures source tracing, and users trusting AI-generated content without verifying facts.
Librarians and educators must collaborate to revamp information literacy for the generative AI era. Here we propose some tracks that center Wikipedia in the conversation:
Web Literacy: Understanding the evolution of web design and information flow, particularly the shift from ""garden"" to ""stream"" models, is essential for contextualizing generative AI’s role in research. This shift affects how knowledge is created, shared, and accessed.
Wikimedia Projects as Models: Wikimedia projects exemplify transparent knowledge creation, offering lessons in content creation, version control, citations, and verifiability. Unlike commercial AI, these projects avoid the tracking-surveillance-attention model, providing a learning-focused web environment free from commercial biases.
Critical Evaluation of GenAI: A module that explores both the limitations and benefits of generative AI. It addresses concerns such as reliability (bias, assumptions, and hallucinations), source transparency, and inclusiveness, while also examining tools like NotebookLM that promote transparency and customization and improve the research process when used responsibly.