Understanding drivers' trust after software malfunctions and cyber intrusions of digital displays in an automated car

Abstract

The aim of this paper is to examine the effect of explicit (i.e., ransomware) and silent (i.e., no turn signals) automation failures on drivers’ reported levels of trust and perception of risk. In a driving simulator study, 38 participants rode in a conditionally automated vehicle in built-up areas and motorways. They all experienced both failures. Not only levels of trust decreased after experiencing the failures, especially after the explicit one, but also some of the scores were low. This could mean cyber-attacks lead to distrust in automated driving, rather than merely decreasing levels of trust. Participants also seemed to differentiate connected driving from automated driving in terms of perception of risk. These results are discussed in the context of cyber intrusions as well as long- and short-term trust.