While flying is certainly safer than driving, that statistic speaks more to the gross volume of cars driven vs. planes flown than anything else.

The perception of danger in planes and self-driving cars is perceived as such due to lack of control. If a plane is about to crash, there's not much to do besides hold and pray. If your car accelerates when it's supposed to brake, you better be paying attention and act fast to avoid a crash. With your hands on the wheel, you're always in a position to act, and it's easier to believe that there's always something you can do to avoid disaster.

These statistics are typically deaths per passenger-mile travelled or similarly normalized. Or do you mean that highway traffic is more dangerous because there are more cars in a smaller space?

Don't forget manufacturer liability. 100x safer self-driving cars would still cause 330 deaths and 22,000 injuries per year. Each one of those could result in a negligence lawsuit against the manufacturer.

Even in the case of "mostly" autonomous vehicles where it could be argued that the human driver was lulled into complacency.

And it's unclear if the insurance industry would shoulder that like they do human errors. Serious accidents can bankrupt responsible people after insurance is exhausted, and those would all now be concentrated on manufacturers with much deeper pockets. Plus there's all the single-vehicle accidents which currently don't result in lawsuits, but would with autonomous vehicles.

I don't think you can ever eliminate all accidents, but the ones that do happen shouldn't be as simple crashing into a truck in front of you because the sensors didn't detect it properly.

This article doesn't mention it but the Tesla press release pointed out... "the vehicle was on a divided highway with Autopilot engaged when a tractor trailer drove across the highway perpendicular to the Model S"[1]

Think about that for a second. How does that happen? Sounds like the truck was jack knifing, making an absurdly abrupt lane change, or the highway was not completed divided. Either way it's not like it was simply rear-ended, the car went under the side of the truck.

I'm not a Tesla apologist by any stretch but this sounds like a bit of a freak thing. My guess is the physical sensor wouldn't pick up something that high off the ground (keep in mind this was the side of the truck not the back where there is a bumper) and any camera sensors didn't have enough contrast. To prevent this they would have to have sensors point UP and forward, which I would guess they will be doing after this.

[1] https://www.teslamotors.com/blog/tragic-loss

tesla press release may not be the most trustworthy source in this case

It's accurate.

http://electrek.co/2016/07/01/understanding-fatal-tesla-acci...

That article is the best one I have seen yet. I like that there is actually a diagram.

There is a bigger tolerance for error if it is your own fault. For example, compare how the risks of active and passive smoking are perceived.