Trust in self-driving vehicles is lower than in human drivers when both drive almost perfectly

Studies have investigated the determinants of trust in self-driving vehicles (SDVs) and confirmed that the ability to execute the driving task flawlessly is essential to promote trust. However, little is known about the extent to which errors decrease trust in SDVs. This study conducted four experiments (N = 2221) and tested whether people's trust in SDVs was lower than that in human drivers when they made errors without causing negative events. In Experiments 1 and 2, which manipulated the driving accuracy of the drivers, the participants checked nine different pieces of information that showed accuracy. The results demonstrated that the SDV was less trusted than humans only when there was a slight possibility of making an error. This study did not identify factors explaining lower trust in the SDV. Experiments 3 and 4 consisted of participants watching videos showing that the SDV and human driver made minor errors, such as taking a long time to park. This study showed that the minor errors largely reduced trust, regardless of whether the vehicle was self-driven or driven by humans. These findings imply that errors without describing severe accidents are less likely to cause a gap in trust between SDVs and human drivers. © 2024 Elsevier Ltd