Looking at the massive downvote of my comment and all the subsequent replies, lots of people here missed their MIT class of Ethics in Software Engineering...
In safety critical engineering, ethics are not opinions but process guarantees.
Waymo system is ML dominant, non deterministic, and validated statistically, without a public end to end safety case, formal failure bounds, or provably safe fallback under unknown conditions.
Shipping anyway is not a technical necessity but a choice to externalize unbounded risk onto non consenting bystanders. Comparing that to bad human drivers or stock prices misses the point that this about what risks you knowingly impose.
Looks like the Waymo Software team could apply at Boeing. I hear they are hiring....
I'd bet all my money, and all the money I could borrow, that a waymo would stop/swerve for a child running out before the sensory nerves in a humans eye reacted to that child. Just thinking it's not as egregious a violation when committed by something with a 0.1ms response time. Still a violation, still shouldn't do it, but the worst case outcome would be much much harder to realize than with a human driver.
Also just to add, the fact that there aren't cases of this from Phoenix or SF seems to signal it's a dumb mistake bug in the "Atlanta" build.
You’re giving a technical answer to a question that’s actually about the economic and policy incentives.
Yes, electronic sensors can enable the car to react more quickly: But react how?
A buggy or unexpected reaction will just lead to equal or faster tragedy.
Individual drivers are incentivized to keep their behavior (or be taken off the road). What legal incentives are there when a faceless company is involved and creates one or two drivers “at scale”?
Your quip about stock options is actually funny, because if the engineers were killing people then those stock options shouldn’t be worth so much.