Will it though? Uber's Pittsburgh fleet alone will be collecting enough data that even .000001% likelihood errors will be coming up in a matter of weeks. And as people always like to point out in this debate, the minimum bar for safety (beating a human driver) is not that high.

I don't think the software is the gating item for full autonomy. The production cycle of the hardware (the actual cars) is what we'll be waiting on most over the next few years.

I could see see something like Uber electric cars / golf carts where driving distances are small such as Sun City in AZ. Baby steps. Can't have a death in an accident though.

Pretty sure 0.01% success rate is still better than human drivers.