Beyond Usefulness: How Cognitive Evaluations and Relational Perceptions Shape ChatGPT Continued Subscription Intention

Abstract

As generative AI adopts subscription models, understanding continued payment intentions is essential. Drawing on the expectation-confirmation model, this study integrates technical quality and human–AI relational perceptions to explain continued subscription intention for ChatGPT. Survey data from 565 paid users show that system and information quality affect subscription intention directly and indirectly through expectation confirmation, perceived usefulness, and satisfaction. Satisfaction has the strongest direct effect. Liking and trust influence subscription intention both directly and through satisfaction, whereas intimacy operates mainly through satisfaction. Perceived threat and anxiety have no significant direct or mediated effects. The findings extend the expectation-confirmation model to generative AI subscriptions and identify satisfaction as the central mechanism connecting cognitive evaluations, positive relational perceptions, and continued payment decisions.