Error interpretation during everyday automation use

Experienced users of everyday technologies may perceive automation errors. This paper examines how those users know an automation error has occurred and with what level of detail users interpret those errors. Thirty participants were interviewed about incidents when their activity trackers made an error. Participants described interpreting those errors to various extents, with causal interpretations being the most frequent, but comprising only 46% of their interpretations. Participants used a variety of cues to understand errors. These cues related to context, measurement comparison, device mental models, checking their device, consistency, component information, and information provided about their device. The types of cues were related to the types of interpretations made by the participants. Additionally, participants sometimes provided multiple interpretations for the same error. Understanding what types of cues promote what types of interpretations is the first step in determining what information to provide to users to help them troubleshoot automation errors efficiently and effectively.