Dynamic Trust Formation in <scp>AI</scp> ‐Enabled Automation: A Longitudinal Case Study of an Advanced Driver Assistance System

ABSTRACT Artificial intelligence (AI)–enabled automation is often utilised in systems that perform human tasks. While useful, these technologies pose a risk of adverse consequences in the case of AI failures as they operate in open environments, learn and are subject to continuous updates. These factors create uncertainty about AI performance that users need to cope with. At the same time, trust in AI‐enabled automation has become an elusive target that continuously evolves. In this study, we focused our investigation on the process through which users continuously form trust towards AI‐enabled automation. Drawing from coping theory, we conducted a case study of Tesla's advanced driver assistance system and qualitatively analysed longitudinal data on real‐life use experiences from a Tesla driver forum. Our findings suggest that users form and evaluate trust continuously as they encounter new experiences and gain new information about AI‐enabled automation, which leads to varying levels of compartmentalised trust and diverging adaptation actions over time. Building on these findings, we elaborate on a process model and change mechanisms of users' trust to provide a better understanding of how trust in AI‐enabled automation evolves over time.