The issue is not the overall track record on safety but how AV accidents almost always involve doing something incredibly stupid that any competent, healthy person would not.

As long as the overall number of injuries/deaths is lower for autonomous vehicles (and as you’ve acknowledged, that does seem to be what the data shows), I don’t care how “stupid” autonomous vehicles’ accidents are. Not to mention that their safety records will only improve as they get more time on the roads.

That’s probably true, but their handling of edge cases will only get better the more time they spend on the roads, and it already looks like they’re significantly safer than humans under normal circumstances, which make up the vast majority of the time spent on the road.

In December, Waymo safety data—based on 7.1 million miles of driverless operations—showed that human drivers are four to seven times more likely to cause injuries than Waymo cars.

From your first article.

Cruise, which is a subsidiary of General Motors, says that its safety record “over five million miles” is better in comparison to human drivers.

From your second.

Your third article doesn’t provide any numbers, but it’s not about fully autonomous vehicles anyway.

In short, if you’re going to claim that their track record is actually worse than humans, you need to provide some actual evidence.

Edit: Here’s a recent New Scientist article claiming that driverless cars “generally demonstrate better safety than human drivers in most scenarios” even though they perform worse in turns, for example.

Exactly. As early as the technology still is, it seems like it’s already orders of magnitude better than human drivers.

I guess the arbitrary/unfeeling impression of driverless car deaths bothers people more than the “it was just an accident” impression of human-caused deaths. Personally, as long as driverless car deaths are significantly rarer than human-caused deaths (and it already seems like they are much, much rarer), I’d rather take the lower chance of dying in a car accident, but that’s just me.

How does “driverless cars hitting people is so incredibly rare that a single instance of it immediately becomes international news” at all signify “boring dystopia”? If anything we should be ecstatic that the technology to eliminate the vast majority of car deaths is so close and seems to be working so well.

Don’t let perfect be the enemy of ridiculously, insanely amazing.

The problem there is that you have to know exactly what you’ve done to mess it up in order to fix the bug, and when I fuck up my system, I usually have no idea what I did.