Quote:
Originally Posted by peter12
Not even that. A human could see a toy truck, and just automatically associate that toy truck with a probability of a child nearby. We don't even know how to program that into a car.
And yes, the above ethical issue is huge. Too bad many manufacturers have said that they will solve it by protecting the consumer of their product. So bye-bye kids, Donnie just bought a Mercedes.
|
I guess it comes down to weighing the likelihood of various scenarios and looking at the benefits of drivers vs. computers. Is a human who can recognize a toy truck but travels about 30 feet before they even touch the brakes (assuming a speed of 45 km/h) going to be safer than an autonomous vehicle which will apply the brakes the moment anything appears in front of it?
And it's not like humans apply any kind of real reason or logic to most emergency driving situations anyway; there simply isn't the time. Plenty of people have killed themselves swerving off of cliffs trying to avoid an animal that would only cause moderate damage to their car if they just kept driving straight and hit the brakes. You could literally program a car to do nothing other than brake or to brake and turn in a random direction when there's an unavoidable accident and it'd still come out safer than a human driver.