Robot Cars

Some of my friends believe that improving technology will solve all our problems, or at least most of them.

One of the technological fixes that many of these friends believe in is automobiles that save energy, and allow greater use of existing transportation infrastructure, by self-driving.

It is true that self-driving technology has improved considerably in recent years.  Though it seemed like one of the tasks that would be solved quickly in the 1950's, and it still seems it has a way to go.  (Basic operations of perception proved far more difficult than computer theorists of the 1950's ever imagined.  We still have a long way to go in understanding language, for example.)

One of my problems with robot cars is what I perceive as a misunderstanding of the problem.  Driving is not merely a technical problem like finding one of the better routes in between points.  Driving is a task that involves constantly judging the assertiveness of of other drivers.  Thus it is more like poker than chess.

Another problem is the liability problem.  I am liable for my own actions, but who is liable for the performance of a robot car?  Well, the robot maker of course, or at least you would think.  But have you ever seen those disclaimers that come with all software?

I suspect that robot cars will only work when all the cars are driven by robots.  And that is pretty hard to imagine.  And when you get there, by the way, what you have is highly computerized segmented train.

Finally, like most technical solutions, robot cars ultimately create the same problems faced with wider roads.  Wider roads don't become less congested, they encourage more people to drive and therefore ultimately cause MORE congestion.