Driving into the Loop: Mapping Automation Bias and Liability Issues for Advanced Driver Assistance Systems

Advanced driver assistance systems (ADAS) are transforming the modern driving experience. Today’s vehicles seem better equipped than ever to augment safety by automating routine driving activities. The assumption appears straightforward: automation will necessarily improve road safety because automation replaces the human driver, thereby reducing human driving errors. But is this truly a straightforward assumption? In our contention, this assumption has potentially dangerous limits. This paper explores how well-understood and well-researched psychological and cognitive phenomena pertaining to human interaction with automation should not be properly labelled as misuse. Framing the problem through an automation bias lens, we argue that such so-called instances of misuse can instead be seen as predictable by-products of specific engineering design choices. We engage empirical data to problematize the assumption that automating driving functions directly leads to increased safety. Our conclusion calls for more transparent testing and safety data on the part of manufacturers, for updated notions of misuse in legal contexts, and for updated driver training regimes.