I've been trying to think this through a bit.
The way I understand it is: There are two cars waiting to go through an intersection. The Cruise and the Human-Driven car. The light turns green, and the Cruise's pathway through the intersection is clear.
A pedestrian is crossing the intersection in the path of the human-driven car, but not in the path of the Cruise.
Both cars enter the intersection, the human-driven car strikes the pedestrian, who then rolls off the human-driven car into the path of the Cruise, and is run-over the Cruise.
Of course, I haven't seen the accident, but if that's broadly correct, I'm not sure that this is the damning evidence of "driverless cars are unsafe" some people seem to be spinning it up to be.
It is a tragedy, but it seems to be a tragedy caused primarily by a human driver who struck a pedestrian, and a human pedestrian who ignored a traffic signal.
Should the bar for driverless cars be "able to predict when other vehicles will hit a pedestrian and roll that person in their path"?
Would a human-driven vhicle have done any better? If the places were reversed, would the Cruise have struck the pedestrian in the first place?
I think we're all right to be skeptical of machine-driven cars, but I'm not certain that this specific incident is a point against them.