Does frequency of use enhance ChatGPT user satisfaction? The role of perceived functional capabilities and AI interaction preference

Abstract

This study proposes a post-adoption socio-technical framework in which perceived functional capabilities of ChatGPT (technical dimension) and AI interaction preference (social dimension) mediate the relationship between ChatGPT usage frequency and user satisfaction. Drawing on continuance theory, expectation-confirmation theory, and socio-technical perspectives on human-technology co-adaptation, we developed a dual-mediation model and tested it using Structural Equation Modeling with data from 12,295 ChatGPT users. The results show that frequency alone is not a reliable predictor of satisfaction once social and technical mediators are considered. Specifically, satisfaction is activated through the progressive discovery of the technical tool’s potential, including cognitive offloading, anthropomorphism, and other task-oriented capabilities, and through users’ social preference for AI-based support over human interaction. The findings contribute to post-adoption and continuance research by offering a parsimonious socio-technical satisfaction activation model, a set of social and technical generative AI-specific activating mediators, and a perspective about the need for ad-hoc complementary technology adoption models for conversational large language model systems. As such, the study provides a foundation for future research to test, refine, and extend the proposed model across domains, as well as to enrich its technical and social constructs with additional context-specific mediators. At the practical level, the findings suggest moving beyond simply encouraging more frequent or responsible use of generative AI, and instead undertaking also complementary actions that address all the drivers of user satisfaction.

Ähnliche Arbeiten