Stephan Weibelzahl

While system performance and user expectations are known to shape user acceptance, their interplay in the context of artificial intelligence (AI) remains underexplored. This study adopts a holistic perspective to examine how user expectations and AI performance influence user experience and acceptance in a concrete interaction scenario. Drawing on the Components model of User Experience (CUE), we conducted a longitudinal experiment (N = 202) in which participants engaged with a mock AI system for emotion recognition, experiencing either high or low system performance. We measured expectations, user experience, and acceptance at three points using validated questionnaires. Results show that user expectations and pre-use attitudes significantly shape both cognitive experience and post-use acceptance. While AI performance enhances cognitive user experience, it has no direct impact on emotional experience or acceptance. These findings highlight the critical role of managing user expectations in the design and deployment of AI systems. Educational interventions and expectation-aligned system features may foster more positive user experience and higher acceptance – independent of actual performance.