Driving (with irrational human actors) is non-deterministic.
I don't know about the rate of deployment, but Google is already lobbying Nevada to officially allow its driverless cars. They have successfully logged hundreds of thousands of miles on U.S. roadways with only one incident: one of the driverless cars was rear ended while waiting at a red light.
> Driving (with irrational human actors) is non-deterministic.
They don't have to "solve" driving. All they have to do is be less stupid than human drivers.
We haven't set the bar that high.
Even so, the Google project seems to be working fine so far, irrational human drivers notwithstanding.
And I bet Google's car is already better then most if not all human drivers can ever be.
The fact that the road is full of unpredictable humans does not mean that a computer can't be a better driver then another human. In fact, the more dangerous the road conditions, due to other bad drivers, or any other reason, the better it is to have a super-human computer do the driving.