From Expectation to Experience: Understanding Public Acceptance of AI-Enabled Autonomous Shuttle Services in Seoul

This study examines public acceptance of autonomous shuttle services in a real-world urban context by integrating expectation–experience dynamics, system characteristics, and configurational analysis. Based on survey data collected from users of Seoul’s self-driving shuttle operating along the Cheonggyecheon corridor (n = 566), a mixed-method approach combining structural equation modeling (SEM) and fuzzy-set qualitative comparative analysis (fsQCA) is employed. The results confirm that pre-use expectations significantly shape post-use experiences, supporting the expectation–confirmation framework. Notably, perceived autonomy exhibits a significant negative effect on user attitudes, suggesting that users may prefer partial automation rather than full autonomy during early deployment stages. In contrast to prior research, trust and satisfaction do not significantly influence attitudes, suggesting a context-specific pattern in which user evaluations may be shaped more by system-related considerations than by psychological responses in this early-stage pilot setting. Furthermore, perceived human backup plays a dual role by enhancing experienced safety while simultaneously reducing perceived autonomy, highlighting a human backup paradox in early-stage deployment. Contextual factors, including integration value and fare acceptability, significantly influence continuation intention, highlighting the importance of system-level integration in public transport. The fsQCA results further uncover multiple configurational pathways leading to high acceptance, demonstrating causal complexity and equifinality. These findings advance understanding of user acceptance in early-stage autonomous mobility systems and provide both practical and policy-relevant insights for designing safe, trustworthy, and system-integrated AI-enabled transport services, thereby supporting the sustainable deployment of autonomous transport systems in smart cities.