Manipulating Rare Print Metadata with ChatGPT

Abstract

This is an account of a small scale trial to evaluate the potential application of AI in the upgrading of existing MARC records, specifically using ChatGPT-4 to render records DCRMR compliant. The trial was informally structured and as such the findings are neither comprehensive nor entirely conclusive. However they do demonstrate the difficulties generative AI has in producing consistent and accurate records, especially with regard to describing the copy-specific elements particular to rare print materials. The principal conclusion of the trial is that while there are effective existing metadata tools to assist in upgrading records, AI may have applications in improving their usability and streamlining the processes involved.