This article looks at how well an AI-assisted literature search works in an area with a limited number of publications (and, specifically, three previously isolated areas of literature and data—Dene kedǝ (North Slavey language), Dene ts'ı̨lı̨ (being Dene, Dene ways of life), and Dene ts'ǫ́ dane (Dene youth)). The tools found a few relevant papers, then filled in the rest with references to papers about different languages. Worse, they missed "thousands of relevant documents about the Sahtú", including digitized archive documents, key academic works, government publications, and sources in physical archives. Why? Faun Rice offers two explanations: first, with a smaller knowledge base, it's harder to determine relevance. And second, using metrics like citation counts tends to sideline less represented voices. I wonder how much if this is unique to AI-based literature reviews, and the formal literature review process in general.
Today: Total: [Direct link] [Share]
