I am reasonably confident that properly designed and coded AI can adequately accommodate the limitless number of unplanned road and traffic scenarios, at least better than many human drivers can.
The programming to accommodate every scenario brings up some interesting ethical questions. I recall listening to a radio show discussion of this topic and the participants were debating the right AI decisions for such things as: “A young lady pushing a stroller steps off the curb in front of the vehicle. The only way to avoid hitting her is to veer right into a bus shelter full of people. Who gets hit?â€
What scares the H@** out of me is the automative industry's laissez-faire attitude toward network security.
Before I will feel comfortable with roads filled with driverless vehicles, I want to see real solid, reliable security technology that will protect these vehicles not only from network vandalism but also protect their user's privacy.
I assume that any vehicle would have a driver over-ride in case the system was hijacked but I certainly agree with your point on user privacy. Hackers would likely have access to things such as your driver’s licence number - a key piece of information in identity theft. Also, a crook would be happy to know you’re out running errands and your house is ripe for a break-in.
