What's the supposed advantage of a driverless robot car?

In reading the answers to this related question , I began to wonder, of the people who think driverless robot cars are a good idea whose time has come, or which will eventually replace cars with human drivers, why they think that?

Even if it were easy to do well (which I don’t think it is) I’m hard-pressed to think of reasons why one would particularly prefer a driverless car. All I can think of are these:

* Some people can’t drive (children, intoxicated people, certain disabled/debilitated people).
* It could be convenient if you’d were someplace without a car, to be able to summon one without paying a lot for a taxi driver.
* It could maybe help some parking and traffic situations (though also create more traffic and use more fuel) to be able to have the car drop people off and not have to park and sit close to where the driver is until the driver is ready to leave. You could also do things like embark a canoe up-river and then have the car come pick you up downstream.
* If many private cars were replaced by public taxis, then there could possibly be less cars needed. Though, this is already possible with human taxi drivers.
* For cargo and ferrying purposes, in some situations you’d have another empty seat, saving the weight and time of a driver.
* Some people don’t like to drive, so this seems like a great convenience to have a robot do it.
* Some people think robot cars would be safer (as a simulation & AI programmer, this seems extremely unlikely to me).
* Some people are appalled at many other human drivers and afraid of accidents and unpredictability, imagine that robot drivers would be better (again, I don’t think so).

I also think of many problems and disadvantages:
* Difficulty of making robot drivers smart enough at understanding conditions to really replace human drivers.
* Gung-ho companies wanting to be first to sell robot cars, deploying them when they are still not really safe.
* Situations where robot drivers think they understand what’s going on, but don’t, so they make bad moves and bad decisions, which would be not just dangerous but terrifying.
* Eventual reduction of number of competent human drivers.
* Computer bugs and failures leading to new types of problems.
* Computer hacking leading to new kinds of crime, assassination, etc (this is already a problem with many modern cars, where a hacker with a wifi laptop could make some cars slam on breaks, turn, or accelerate). Or, hackers hit central traffic database and give robot cars false data about where traffic or even roads are. Computer virus spreading to cars could immobilize many cars.
* What are these AI’s going to do in human situations that an AI isn’t going to be able to understand, such as road rage, health emergencies, or criminal activity?
* I and others like driving cars ourselves, and would not like or trust a robot car.
* Part of driving safely is watching other drivers and predicting what they will or will not do, and communicating by looking at the drivers and gesturing, etc.
* As an Artificial Intelligence (AI) programmer myself, I can’t believe anyone thinks (correctly) that it’s a good idea to replace all of the decision-making tasks of a human driver with an AI. Not only would it be a ridiculously complex task to even make an AI that can understand all the situations that can occur, but to have a program making decisions which can mean life or death, seems not only insanely (probably impossibly) difficult (since they can be debatable even for humans sitting down and analyzing a situation after the fact), but undesirable to have a computer making those decisions rather than human drivers.
* And then there are the legal implications, such as the one in this question .

To me, this adds up to “not ready for prime time, would be dangerous, and even when perfected, it only seems desirable for some circumstances, not all or even most.”

So I wonder, for those who think it’s coming soon and/or eventual, am I missing some advantages?