While human drivers can only react instinctively in a sudden emergency, a robot car is driven by software, constantly scanning its environment with unblinking sensors and able to perform many calculations before we’re even aware of danger. They can make split-second choices to optimize crashes–that is, to minimize harm. But software needs to be programmed, and it is unclear how to do that for the hard cases.Ugh, I can't believe I didn't consider this challenge for google's autonomous car programmers back when I was posting stuff like this dial-twisted version of the dilemma.
A small child darts into the highway, there's no way to stop in time. Should the autonomous car carrying a family of four with two small children veer left into oncoming traffic? Veer right where a police car has pulled over a van the child just escaped from? The driver of the van is a known serial killer and the vehicle's facial recognition program has matched him with 92% certainty to Most Wanted list, but veering right would also mean killing the decorated police officer and the infant she is holding ... but there's no way to know if the infant is actually a Hitler clone baby ...
Link via Gerry Canavan