I have written before about why I think driverless cars should never be allowed, but this time I’m going to suggest why the public — you — would likely reject the idea.

During the transition to a driverless car, you’ll get to experience, as I already do, some of the features required for a car to drive itself. Those features include traffic-aware cruise control and lane management dependent on multiple cameras, radar and numerous sensors. I have been using those features on my Tesla for quite a while.
The first thing to recognize is that a self-driving car will always err on the side of caution. Here are just three examples: Let’s say you’re driving a city street with parked cars but no line between the travel lane and the parking lane. Every now and then your car will mistake a parked car for a stopped car and simply stop.
Or you’ll be driving along and a car coming the other direction with make a left turn in front of you. Erring on the side of caution, your car will abruptly apply the brakes even though it’s clear to you that braking was not needed.

Or you’re driving on a road with no bike lane, but there’s a cyclist cruising along at 10 mph and no room to pass without crossing the yellow line, which your car won’t do. You car slows to 10 mph.
My Tesla knows the speed limit on all roads based on GPS information, but 1) sometimes the GPS information is wrong, and 2) sometimes there’s a lower speed limit in effect for school zones or construction. Your self-driving car will plow through those areas, totally oblivious!
Wildlife poses a special problem. As a human, you know to slow down if an unpredictable deer is next to the roadway. You driverless car doesn’t have that judgment.
Think of all the times you depended on exchanging eye contact or body language with another driver to know whether to yield or not yield. Think about two lanes merging into one, or about another car being driven erratically. Think about going off road. Think about anything other than driving on a dedicated highway with other driverless cars.
Think about seeing someone in distress on the side of the road or within sight of you. Think about witnessing an accident. Your car will want to leave the scene of the accident rather than stop.
Think of when the painted lanes have disappeared due to wear and only a human could figure out where to go. Or lines that have not been removed completely when new lines were painted.

In Golden, where Hwy 6 crosses Colfax Avenue, it’s not a 90-degree intersection. If I’m in the left lane traveling west through that intersection, my Tesla consistently misinterprets the dashed guide lines for the left-turn lane next to me and swerves into eastbound traffic thinking that it’s a left curve. Fortunately, I have my hand on the wheel and make the immediate correction.
I hope by now you have gotten the impression that self-driving software can not anticipate every conceivable (or inconceivable) situation and could lead a driverless car into desparate situations.
Hi, Jim. Not there yet, I agree (I think my Tesla drives a bit like a crazy teenager) — but never say never. When the car’s AI results in substantially fewer deaths than human-caused mistakes, it will make sense. Might be 5-20 years (just guessing, based on my background in AI). Just as AI has eventually overtaken human capabilities in, e.g., Chess, Go, etc., I see no reason it won’t happen. Right now the AI is in the “learning” phase, figuring out how to drive better, getting a handle on all the outlying cases (such as you mention). There are lots of reasons why people will want it, too. (More time to do other things while driving, it can drive you home if you’re drunk, etc.) I’m looking forward to it.
LikeLike
What you may not be considering is the psychological/public relations component. The public isn’t driven by rational facts as much as emotion. (Trump supporters demonstrate that!) Yes, if the public is willing to put up with news reports of “stupid deaths” (one’s that wouldn’t have happened if someone was watching) long enough, they MIGHT be convinced logically, but just a few more of those stupid deaths (which you can be sure will be well publicized) and the public will turn against the concept. Also, the majority of people do NOT drive drunk, and that argument of surrendering a potentially death-causing activity to a computer will not appeal to the person (like me and perhaps you) who has never even had more than a fender bender.
LikeLike