Language expertise, or lack thereof, is not a new problem in cataloging departments. Libraries often collect in languages that no on-site cataloging staff can read or speak. Languages that require non-Roman scripts also require transliteration of characters as well as the ability to translate enough of the title and publication information to record an accurate description. LLMs absolutely have the potential to allow faster and easier creation of MARC records, with some degree of expert human oversight. This presentation will share the results of work by two librarians who are working to develop methods for using Large Language Models (LLMs) for MARC cataloging of Chinese language materials. We look at the ability of an LLM to provide Chinese characters and Pinyin transliteration, as well as a complete MARC record. While this case study looks at Chinese language materials, the methods could be applied to other languages that use non-Roman scripts as well. This presentation will discuss development and use of prompts and give examples of results generated by ChatGPT or other LLMs. Participants will get a feel for the overall benefits and drawbacks that come from working with AI and learn about tools that may be useful in their work.
