Those self-driving driverless cars scare me. Let's say you're driving a car, and someone else runs a red light or is drunk and driving crazy, and that other car hits your car. As your car careens off course, you attempt to take control and several options present themselves to you.
One path will have you running over a young mother pushing a baby stroller. Another path will have you running over several small children. Another path will have you smash into a bus stop with a homeless guy on a bench. You have limited control and these are your options.
Can a computerized self-driving car make the most humane decision as to whose lives to spare? I've seen collisions like this happen within my sight, with a hit car deftly avoiding spinning into pedestrians in a crosswalk, just barely. Can a computerized driver do the same?
EXACTLY!
I just as soon neither be IN one nor on the same road as one...
The computerized car will never have that problem.
You see, at Level 5 autonomy, cars will talk to each other, completely avoiding collisions with each other. Vehicles do not talk to a central network for this, just to each other.
The computer can make adjustments far earlier than a human can possibly make those same adjustments.
The drunk drivers, inattentive drivers and poor drivers will make autonomous vehicles happen faster because it gets those people off the roads first.
With 95% of vehicle fatalities caused by human error, there will be no argument against autonomous vehicles that makes sense.
Driving and car ownership will be reserved for the wealthy who want to race on a track or course, just like horses are today.
There are currently ~40,000 traffic fatalities per year. If that could be reduced to ~15,000 does it really matter that much to you that the most humane decision is made in each collision? In other words, what is your metric?
I would say total traffic fatalities would be one very reasonable metric, and if autonomous vehicles could reduce this significantly (e.g. more than 20%, and statistically significant relative to year-to-year variance), then that is a win. Second metric should be average trip time; also a 20% significant reduction in point-to-point travel would be a win. Third, reliability and maintenance cost-- should retain reliability and lifetime cost of ownership should not increase by more than 20%. Given benefits of being able to sleep, or do other things while driving, if autonomous vehicles could also do these three things, it would be an amazing change. All of these metrics would depend of course on having the majority of vehicles on the road be autonomous-- you would have to extrapolate available data to intermediate metrics to see if we are trending towards these goals initially.