Perhaps the issue here is that Autopilot and active safety features are more likely to be activated in highway driving rather than city driving which is messier

Perhaps the issue here is that Autopilot and active safety features are more likely to be activated in highway driving rather than city driving which is messier. These stats might look quite different if the metric was accidents per hour of driving rather than accidents per mile of driving. In addition, most people's safety concerns are more about how safe the car is when it's in an accident. A car accident you walk away from unhurt is an inconvenience whereas an accident in which you're injured can have lasting quality of life implications.