“Domo arigato, Mr. Roboto”: A qualitative content analysis of AI music in WorldCat

Abstract

Artificial intelligence (AI) has been used for experimentation in generating music for the last seventy years, but recent advances in generative AI (genAI) have led to novel, creative, and even surprising results. Issues arise when genAI and human efforts are simultaneously recognized in a creative work, constituting the uncanny valley and leading to discomfort among listeners. Additionally, the lack of transparency required of media producers regarding genAI use robs audiences of their right to choose whether to engage or avoid genAI content. This has sparked discussions among researchers, industry leaders, and lawmakers about regulating genAI use with priority given to enforcing transparency. Libraries can play a role in this by curating metadata when cataloging genAI materials, but current cataloging practices and policies inhibit the cataloger's ability to maximize accuracy and transparency when describing genAI items. This study features a content analysis that examines WorldCat item records belonging to genAI songs and music albums and finds inconsistent item record descriptions, often vaguely referring to or omitting genAI use. Supported by epidata theory, this study recommends adopting Resource Description and Access (RDA) standards to improve accuracy and transparency in cataloging genAI music.