The first step in every Autopilot crash is always the same: the system works, fostering the perception that it's more capable than it is and increasing driver complacency... 
Step two is typically consistent as well: the system encounters something that the system isn't able to handle and the driver has become complacent, leading to tragic consequences.
In other words, Tesla's attempt to frame Autopilot safety as a simple comparison of human vs automated safety is a red herring. Autopilot is the "uncanny valley" of automation. Screencaps via this simple yet informative video. bbc.com/news/technolog…
