If you want the technical reasons why self driving cars will have practically insurmountable difficulties in the real world, I would recommend Artificial Unintellgence by Meredith Broussard. Excellent resource.
But I don’t think the technical challenge will be the nail in the coffin. It will be people’s feelings.
Humans don’t trust robot cars, or the motives of those building them, and have been attacking them with rocks and knives according to the NY Times:
“They said they need real-world examples, but I don’t want to be their real-world mistake,” said Mr. O’Polka, who runs his own company providing information technology to small businesses.
This is just the beginning. Those who think they can win this argument by saying “but robot cars are safer” are missing the point. Lots of safer alternatives in American life are rejected because of the perceived loss of freedom and control. It is safer to not jaywalk, but visit New York sometime. It is safer to not ride horses, skateboards, and motorcycles. It is safer to use a computer not connected to the internet.
The thing that will stop the self driving car will be laws that require that robot actions be backed by humans. Humans who must make moral choices, who must observe.
Driver assist will happen. Machine learning will make driving safer. But jumping to self driving cars is like suggesting that an AI could run a city government, and trying it out on real citizens, before the algorithm can even manage a MacDonald’s successfully first.
As cities begin to pass regulations against self driving vehicles I will detail them on this blog.
Painting by Simon Stålenhag