What influences the continued use of generative artificial intelligence: Evidence from covariance-based structural equation modeling and fuzzy set qualitative comparative analysis

The net effect of a single variable is insufficient to explain the causal complexity underlying users’ continued intention to use generative artificial intelligence. To address this limitation, this study further investigates configurational effects to overcome the shortcomings of traditional linear and symmetric models. Drawing on information ecology theory, the unified theory of acceptance and use of technology model, the DeLone and McLean model of IS success model, and configuration theory, this study applies covariance-based structural equation modeling and fuzzy set qualitative comparative analysis to analyze 291 valid questionnaires collected from university students in Hangzhou, China. The covariance-based structural equation modeling results show that effort expectancy, social influence, service quality, system quality, and information quality exert significant positive effects on users’ continued intention to use generative artificial intelligence, whereas the positive effect of performance expectancy is not significant. The fuzzy set qualitative comparative analysis results indicate that high continued intention to use generative artificial intelligence arises from the combined effects of multiple variables across four dimensions, namely information user, information environment, information technology, and information, under different configurations, with multiple configurational paths leading to the same outcome. These findings move beyond the limitations of traditional “one-size-fits-all” approaches, provide practical insights for generative artificial intelligence developers and operators in product optimization and operation, and contribute to the education and application of artificial intelligence in universities.