Avoiding a car accident can be a matter of split-second decision making and perfect timing. Sometimes, even the most attentive and skilled driver will be put in a position where an accident is unavoidable. As you approach an intersection, a car runs a red light and you are forced to swerve. You have cars on one side and a pedestrian on the other. Someone is going to get hit. The safest course for you, in your car, is likely to hit the pedestrian. But the chances of that accident causing a fatality are higher than in swerving into cars or going forward and hitting the car running the red light. You will likely make your choice by reflex, without any real contemplation.
Autonomous vehicles don’t have reflexes. They must be programmed to perform a certain way. If they are programmed to protect their occupants, the loser will usually be the pedestrian. That ethical dilemma is just one of the things that must be overcome before self-driving cars are made generally available to consumers.
The vast majority of car and truck accidents that lead to injuries are the result of human error. Autonomous vehicles could save thousands of lives a year. But even among the strongest supporters of autonomous cars, you’ll find few who think the accident count will reach zero. Hardware and software defects will still occur and the results could be deadly. Who is at fault in such an accident? How is the victim compensated? How many failures are too many?
It will likely take years for self-driving cars to be a viable option for consumers. In the meantime, more and more vehicles are coming equipped with accident avoidance, auto-braking and other features that incorporate some element of autonomous operation. Drivers may have some time to get used to their cars taking control before they hand over the reins fully.
Source: Reuters, “Look, no hands! Test driving a Google car” by Paul Ingrassia, 17 August 2014