Wednesday, August 10, 2016

Maybe Don't Call It Autopilot?

I'm actually not all that much interested in the safety issues of self-driving cars. It's in a way the easiest part, though really just a subset of the "these things won't work" issue. If self-driving cars really work as promised, they'll be pretty safe. Until then, they won't be both safe and useful. They're sort of one and the same. The focus on safety is going to obscure the more general issue. Won't be safe until they work, and they won't work so they won't be safe.

And every time a crash happens, cultists advocates will come forward to tell us that they're safer than regular cars and besides that the crash was really the driver's fault because autopilot doesn't really mean autopilot so the driver has to know when to pay attention. Maybe all true! Also, misses the point.

If the driver has to pay attention it isn't a self-driving car. And the self-driving cars are never going to happen (in my lifetime, yes, yes, one day our descendants might upload their brains into self-driving car bodies). Things which are a bit more self-driving but are really just cruise control plus will become more widespread and the technology will improve. They still won't be self-driving cars. They'll both be the greatest thing ever and anyone who actually uses them as advertised (implied, at least) will be mocked for daring to do what the technology supposedly promised (of course the car crashed! the driver wasn't paying attention! it's the driver's fault).

The technology will never fail, it can only be failed. Maybe you'll like your new toys, but they won't be self-driving cars.