It is widely anticipated in the media that autonomous vehicles (AVs) will be safer than human drivers. From the National Motor Vehicle Crash Causation Survey (NMVCCS) database, it has been estimated that the final failure in the causal chain of events in 94% of crashes is due to human error. There is the expectation that removing the human driver from the vehicle will eliminate most of these crashes, however this is not a guarantee because AVs will still crash if they are not programmed to avoid the errors that humans make today. The purpose of this presentation is to highlight the types of crashes that may still occur in an all-AV fleet if AVs are not designed to avoid poor choices that currently lead to crashes. Our analysis of the crashes in the NMVCCS database shows the interconnected complex roles that AVs will have to perform and the risks that could arise from rider preferences that AV designers and regulators will need to address if AVs are to maximize their safety potential.