Virginia tech university libraries will investigate how artificial intelligence-generated metadata in digital libraries is represented

The Institute of Museum and Library Services awarded Virginia Polytechnic Institute & State University $946,806 on September 1, 2026, under the National Leadership Grants program (CFDA 45.312) to investigate how artificial intelligence-generated metadata functions in digital library infrastructure.

Virginia Tech's Libraries will conduct empirical research to determine whether large-scale machine-generated metadata can reliably serve as structural infrastructure in patron-facing digital library systems or whether its variability introduces limitations that must be understood before incorporation. The project will identify which categories of AI-generated metadata maintain sufficient consistency when produced across large corpora, analyze structural relationships that emerge across collections, and evaluate how these relationships influence discovery and interaction with digital materials. The deliverables will establish research-based evidence to guide libraries in determining whether and how large-scale AI-generated metadata can be incorporated into digital library infrastructure while upholding principles of reliability, transparency, and long-term stewardship.

Work is performed in Blacksburg, Virginia. The period of performance runs from September 1, 2026, through August 31, 2029.

The project seeks to determine whether large-scale machine-generated metadata can serve as reliable infrastructure or its variability introduces limitations that must be understood and overcome before IT is incorporated into patron-facing digital library systems. The intended project results are to establish empirical evidence about when machine-generated metadata can function as a structural component of digital library systems and when IT cannot. Specifically, the project will identify which categories of AI-generated metadata remain sufficiently consistent when produced across large corpora, analyze the structural relationships that emerge across collections, and evaluate how these relationships influence discovery and interaction with digital materials. The goal is to provide libraries with a research-based foundation for determining whether and how large-scale AI-generated metadata can be incorporated into digital library infrastructure without undermining the principles of reliability, transparency, and long-term stewardship that govern professional practice.