Close

Presentation

How Does Variation in AI Performance Affect Trust in AI-Infused Systems: A Case Study With In-Vehicle Voice Control Systems
DescriptionThe frequent updates of artificial intelligence (AI) infused systems raise concerns about the unstable performance of the systems, due to uncertainty and inexplicability characteristics of AIs. The variation in the system performance may affect users’ trust in the system and their reliance on the system. Specifically, for systems that have relatively high failure rates but low-risk outcomes, for example, voice control systems (VCSs), users’ trust in the system might be based on their interactions with the systems over a relatively long period. In this study, an experiment was conducted with 27 participants to investigate how variations in VCS performance affect users’ trust in it. We found that how a system evolves impacted users’ trust in the system, but the influence of earlier system versions faded out. Further, users’ perception of the system performance had a stronger influence on users’ trust in the system than the actual performance of the system.
Event Type
Lecture
TimeTuesday, September 10th9:45am - 10:05am MST
LocationFLW Salon D
Tracks
Usability and System Evaluation