The Ethics of Abuse and Unintended Consequences for Autonomous Vehicles

Abstract

Abstract There are several foreseeable problems of abuse for the advent of autonomous vehicles (AVs), which will have risky and often unintended consequences. One is that abusive drivers might play “chicken” with AVs: if they know that a car in the oncoming lane is autonomous, and will swerve to avoid a collision, they could purposefully steer directly at the AV in order to cause trouble. In perhaps the worst imagined cases, drivers could play chicken on narrow roads or cliff sides, where an AV swerving could mean serious injury or death for its passengers. Solutions to the “chicken problem” will involve countermeasures ranging from taking photographic evidence (which would not prevent the casualties, merely document them) to making AV behavior less predictable, or even allowing AVs to deliberately collide with drivers playing chicken. Any such programming solution, of course, also has its costs, as the values of transparency and the autonomy and safety of drivers may conflict with such proposed solutions. More generally, whether it is playing chicken or hacking back against V2V cyberattacks, AVs may need to adopt offensive countermeasures against abusive humans. But humans could modify their behavior in response. New norms of driving could emerge. This chapter concludes by examining the likelihood of a process of parallel evolution between AV programming and human behavior that calls for continuous reexamination and modification.