And the systems fail in ways that a human wouldn’t which makes it more dangerous. Teslas often “shadow brake” when driving under overpasses or overhead signs because the radar system gets confused.
This is what gets me. A big part driving safety is defensive, ie monitoring the other drivers around you. It's relatively easy to pick out distracted, erratic or risk-taking drivers from a handful of simple queues. Very much not the case with software driven autopilot routines from an increasing number of vendors. So accidents may go down where humans are at fault, but I find it more unsettling overall, as I have less control to take acceptable safety precautions.
