I'm all for any "assist" systems that use computer speed to reduce lag of human reactions... or sensor perception processing that is more capable that what humans can do (e.g. see in the dark). These safety assist systems should be conservative (overly cautious) so that it's OK to occasionally nail the brakes hard for a piece of garbage blowing across the road. The safety assists should never be thought of as infallible by the attentive operator.
Safety assists should never be packaged and sold or enabled only with "self driving" features, I'm totally against that.
Problem arises when the guy behind you does not have the same level of assist or human capability, and his car ends up piling into the rear of your car. However, rules are pretty clear that's always his fault for traveling too close... whether it was you or assists doing the stopping, won't matter. You preventing an injury accident (e.g. striking a pedestrian) and taking one for the team (stopping all traffic behind you from doing the same by putting mass between the pedestrian and additional traffic is a clear win for the pedestrian. Everyone else in the pileup had a cage around them... and have better odds of surviving impacts.. the cars are completely replaceable and matter not.
Safe full self-driving cars are a fallacy on our current roadways. It's not about the car. It's about our infrastructure (the road) that is not capable of supporting it.
When will transportation systems (including vehicle manufacturers) pony up and invest in making some safe stretches of road that are capable of supporting self driving cars? It's a complete re-think of Musk proportion to tackle the issues. Maybe divert some energy and resources (divert the Boring Company to this "boring" problem of roads) rather than duck it (literally)!