SC Connecticut News

Yale Library has unveiled a prototype application, "Digital Collections AI," designed to leverage Artificial Intelligence (AI) for research within its digitized collections. This tool uses AI-powered Large Language Models (LLMs) to swiftly process, summarize, and analyze digital texts transcribed with Optical Character Recognition (OCR), then answers questions about the content.

The development of Digital Collections AI is spearheaded by Michael Appleby, director of software engineering in Library IT. Appleby initiated this project while experimenting with AI for a conference presentation. He is now collaborating with faculty and graduate students for beta testing.

Reflecting on his time as a graduate student at Yale, Appleby stated: “I hope that this application will help students and researchers to unlock the secrets of our collections and more easily identify the resources they need.”

Jonathan Manton, director of Digital Special Collections and Access at the Beinecke Rare Book and Manuscript Library, sees potential in the tool: “It can allow you to ask questions, say, of a handwritten manuscript with hundreds of pages... It can also provide you with summaries of the text as well as a stylistic analysis.”

Appleby demonstrated the tool using Mrs. E. A. Kenah's travel journal from the 1820s. The tool categorized and summarized the journal's text and responded to specific prompts created by Appleby.

The application allows users to apply different LLMs like Gemini, Claude, or ChatGPT on the same text for varied insights. For example, Claude provided key point summaries while ChatGPT offered detailed descriptions.

Currently limited in language capabilities primarily to Latin scripts such as English and French, challenges remain including correcting poor OCR scans and addressing LLMs' tendencies to generate inaccurate information.

Yale Library seeks collaboration from faculty and graduate students interested in testing Digital Collections AI. Funding is available for research assistants who wish to incorporate it into academic activities.

Lauren Di Monte commented: “We are envisioning that interested instructors might use the tool to support their students’ engagement with digitized special collections or...learn more about how LLMs work.”

Instructors wishing to collaborate or inquire about funding should contact Lauren Di Monte for further information.