I just spent about 15 minutes reading through some of the comments on this Gizmodo post about why self-driving cars shouldn’t have steering wheels. The basic principle is that passengers in self-driving cars aren’t paying attention and don’t expect to be driving, so giving them a steering wheel is probably more dangerous than not giving them one. After all, if you’re always going to be ready to take over the controls of a car yourself, then you might as well just get a regular old human-driven car, right?
The objections in the comments nearly all seem to be on the order of “what if the car fails or encounters unforeseen circumstances?” The answer is, of course, that the car will be programmed to deal with adverse conditions in the safest possible manner. Presumably sensors will determine if the car loses traction on black ice, encounters poor visibility, etc., and slow down or stop the vehicle. If the brakes or steering systems fail then it’s possible that the car could crash. However, this would be true in human-driven vehicles as well, and not all humans know the safest way to handle these situations.
One comment that I saw repeatedly (in several different versions) is that if the car broke down the user would be “stuck.” But this once again assumes that somehow computer-driven cars would break down more frequently than human-driven cars. I’m not sure that’s true.
More importantly, in a world of self-driving cars, breaking down would actually be much less of a problem than in the current world. It would be easy to get another self-driving car to pick you up if you got stuck somewhere. Indeed, Google’s plan for the cars is not that individuals would own them, but that they’d subscribe to a car service. If one car broke down, it could immediately summon a replacement in a matter of minutes. The passenger would change cars, and needn’t concern herself with repairs for the Google car; that’s not her problem, it’s Google’s!
Self-driving cars under this model solve a lot of current problems. There wouldn’t be nearly the same need for public parking; once you’re dropped off at a location the car moves along to the next passenger. Homes could be smaller because there would be no need for garages. Studies have shown that much urban traffic congestion is caused by people searching for parking spaces; this would be unnecessary in a world filled with autonomous vehicles.
It’s not to say that autonomous cars wouldn’t have problems. I’m certain they would; there are issues with hacking, privacy, and how the cars would handle peak travel periods. Would prices go up? Would it take an unacceptably long time to get a car during rush hour? And as I’ve mentioned before, there’s also the human cost: not so much in terms of the physical safety of the vehicles, but in terms of all those taxi, truck, and bus drivers who’d be put out of work.
It’s going to be a much tougher challenge figuring out how to live in a world where robots do most of the work than it will be figuring out how to build the robots themselves.