There’s an old familiar moral twister that used to be taught in school:
You see a trolley moving towards 5 people lying on the track. You can pull a lever to redirect the trolley and save the 5 people, but there is 1 person on the side track that would be killed. What do you do?
Luckily, these days we don’t have to make this grueling decision in our day to day life. However, our generation could soon be faced with a similar quandary. Instead of a train with people tied to the tracks, imagine a car and people on the crosswalk. You have to decide if you keep going straight and hit 5 people or if you should swerve to miss them and hit 1 person. What do you do? What if the 1 person is a child? If I’m being honest, I’m not the best driver and I would probably panic and end up taking all 6 somehow. In reality, there probably isn’t even enough time to make this decision. Everything happens so fast, fate will make the choice for you. We are only human.
But what if cars aren’t limited to human capabilities? The split-second decision is too fast for a human in the moment but it is possible for a computer. When cars are programmed to be able to process these sorts of decisions, how will we handle that? How do we solve the Trolley Problem? What about the human who is programming the computer to make these decisions 20 or even 30 years from now? Should they direct the technology to minimize fatalities? Should they do so at the driver’s expense if it comes to it?
Autonomous vehicles have received quite the examination in the public eye the past few years. We keep hearing the technology is here but there are regulations and other hurdles still to be cleared, including the framework for a “Trolley Problem” incident. These emerging technologies bring questions but also revolutionary possibilities that are both scary and exciting. Many regulatory authorities and insurance companies warn of the dangers of adopting these technologies too early. The primary objective being the cost of the unknown risk. While this is a valid concern, a major point that is ignored in this argument is the cost of waiting to adopt these technologies.
Researchers at McKinsey & Co. have predicted that autonomous vehicles will be able to decrease motor vehicle fatalities by 90%. If that’s truly the number, how many lives are we sacrificing while we wait to “figure this all out”. Did other emerging technologies wait on the politics and regulations to be all worked out before they went into play? What about drones? Why was the climate so different when these sorts of technologies came out for planes?
Humans certainly don’t have a perfect driving record, so what standards should we hold computers to? Even if it’s not perfect, no one person or machine is going to be able to solve the trolley problem. But it seems we are getting a hell of a lot closer.