What drives continuous use of AI-generated music? Insights from a hybrid SEM–ANN approach

AI-generated music is lowering barriers to creation and reshaping music production. As AI music-generation platforms become increasingly capable and human-like, understanding the factors associated with sustained user engagement is increasingly important. Anthropomorphic characteristics may evoke admiration and wonder but may also evoke threat and unease. Integrating the stimulus–organism–response framework with cognitive appraisal theory, this study examines the relationships between anthropomorphic characteristics and continuous use intention through positive and threatened awe and assesses whether these relationships vary according to AI literacy. Survey data from 435 users in China were analysed using partial least squares structural equation modelling and PROCESS, complemented by artificial neural networks. All 11 hypotheses were supported. Anthropomorphic characteristics were positively associated with positive awe and negatively associated with threatened awe. Positive awe was positively associated with continuous use intention, whereas threatened awe was negatively associated with it. AI literacy strengthened the positive relationship between anthropomorphic characteristics and positive awe and attenuated the negative relationship between anthropomorphic characteristics and threatened awe. The moderated mediation indices were also statistically significant. The structural model explained 44.8% of the variance in continuous use intention, while ANN analysis assigned the highest normalised importance to threatened awe in predicting continuous use intention, followed by positive awe and anthropomorphic characteristics. These findings suggest that anthropomorphic characteristics are related to continued engagement through contrasting emotional pathways and that AI literacy conditions how users appraise AI music-generation platforms.