The ethics of driverless cars

Aug 21, 2014 by Andrew Stokes
The ethics of driverless cars
Credit: Craig Berry

Jason Millar, a PhD Candidate in the Department of Philosophy, spends a lot of time thinking about driverless cars. Though you aren't likely to be able to buy them for 10 years, he says there are a number of ethical problems that need to be tackled before they go mainstream.

"This isn't an issue for the next generation, it's happening right now. Driverless cars are on the road in certain jurisdictions as they're being prepared for a ," says Millar, whose dissertation focuses on robot ethics and the implications of increasingly autonomous machinery. "These cars promise safety benefits, but I'm interested in what happens to the cars in a difficult situation, one where lives are on the line."

To explore this problem he created a thought experiment, called the Tunnel Problem, which attracted hundreds of thousands of readers and commenters online. The Tunnel Problem reworks ethical philosophy's Trolley Problem.

The setup is this: You are driving in an along a narrow road, headed towards a one-lane tunnel when a child errantly runs on to the road and trips. The car cannot brake fast enough to avoid hitting the child and so it must decide whether to swerve off the road, effectively harming you, or remain driving straight, harming the child.

"This is a problem with only bad outcomes that even a human driver cannot easily solve," says Mr. Millar. "What's particularly useful about this situation is that it focuses our attention on a design question, as the car will be programmed to respond a certain way—I want to ask who should make the decision about the car's response."

After initially posting his article on Robohub.org, the site ran a poll to gauge readers' responses and rationales as to who should render the judgement.

"A near majority responded that the passenger in the car should have the right to make the decision about whether to swerve or not, and only about 12 per cent suggested it should be up to the car's designers," he says. A full third of respondents said it should be left up to lawmakers and legislators to make the call.

"That so many people were willing to trust a life and death situation to politicians and lawmakers really surprised me," Mr. Millar says. "Many of them said they wanted a standard behaviour so that people would know what to expect in that situation, while others simply wanted someone else to make the decision and take it off their hands."

The Tunnel Problem is one of just a series of problems that Millar foresees being an issue with driverless cars. "There's also the problem of who's culpable when a car crashes. If we maintain current standards of product liability, then the fault will tend to lie with the manufacturer, but we may also shift to a system where we consider the robot at fault," he says.

It's a possibility, but Millar says the future of is far from certain. "Holding the robot responsible may be less satisfying for those with a mind for punitive justice."

Explore further: Should your driverless car kill you to save a child's life?

add to favorites email to friend print save as pdf

Related Stories

Audi tests its A7 driverless vehicle on Florida highway

Jul 29, 2014

German automaker Audi made use of a Florida law passed in 2012 that allows for testing driverless vehicles on Florida highways this past Sunday and Monday, by requesting a shutdown of Tampa's Lee Roy Selmon ...

Really smart cars are ready to take the wheel

Jul 17, 2014

Why waste your time looking for a place to park when your car can do it for you? An idea that was pure science fiction only a few years ago is becoming reality thanks to automatic robot cars.

Cruise aims to bring driverless tech to life in 2015

Jun 24, 2014

The old saying why reinvent the wheel will resonate with the coming debut of Cruise technology in certain cars on certain roads next year. The motivating question would be, Why wait to buy a totally driverless ...

Recommended for you

Google to test cars without a driver

Sep 16, 2014

Google plans to begin testing its new prototype of a self-driving car - which, unlike earlier models, doesn't require a back-up driver - at NASA's Ames Research Center, just a few miles from the tech company's ...

Self-driving cars now need a permit in California

Sep 16, 2014

Computer-driven cars have been testing their skills on California roads for more than four years—but until now, the Department of Motor Vehicles wasn't sure just how many were rolling around.

User comments : 6

Adjust slider to filter visible comments by rank

Display comments: newest first

michelle_rosner
1 / 5 (2) Aug 21, 2014
This is the scenario to solve? "The car cannot brake fast enough to avoid hitting the child and so it must decide whether to swerve off the road, effectively harming you, or remain driving straight, harming the child."

I'm surprised that users had a problem answering that question. Irregardless of situation, I would always want the car to swerve and avoid hitting the child. Problem solved.
foolspoo
5 / 5 (1) Aug 21, 2014
You may be one of the few that has no issues believing your version of morality is the only correct one, Michelle. but we are all surprised that you believe "irregardless" to be a word.
ryggesogn2
1 / 5 (2) Aug 21, 2014
Driver-less cars will solve a major problem for those who need to load the car with explosives and find drivers.
shialar
1 / 5 (1) Aug 21, 2014
So, the question is whether to nearly guarantee the death of the pedestrian, or to risk the passenger being injured to some degree (possibly none)? That actually sounds like an easy math problem/ decision to me.

"A near majority responded that the passenger in the car should have the right to make the decision..." - if the decision is being left to the passenger, then the decision won't be able to happen fast enough to be effective. Thus a decision of inaction was already chosen.
Zera
5 / 5 (1) Aug 21, 2014
Well I know I'm not going to be popular for this response, however, I shall press on.

1, What behaviour has resulted in the passenger being endagered? Answer: Nothing.

2, Why is a child on the road? Answer: Disobeying the rules by which society lives (note i say live here, because the child won't be for long).

3, What behaviour should the self driving car conduct? Answer: A hierachal anaylsis of the situation, if the variables dictate breaking will be ineffective in collision avoidance, is there a viable pattern of movement, determined by physicists, engineers, drivers, and other content experts that will see a zero harm scenario? If the scenario is truly to result in death, let it be the child who is stupid enough to run out on the road (the parents obviously didn't raise it right), if Darwin were alive he'd agree with me.

We're straying towards Asimovs Laws: (http://en.wikiped...obotics)
1a, A robot may not deem one life of more import than another.
veritas1989
1 / 5 (1) Aug 22, 2014
There can be several answeres to the solution.

1. Easiest decision in this situation would be to mimic Human behavior. The robot detects a pat situation where the autcome is always false, so he uses the patern a Human would. Reflex! Given the estimated time, the human would never be able to make a ethical decision, he would react.

2. If enough data is present, and sensoric good enough to destinguish between dog and child the robot could make a decision, on possible survival propability. (like in the movie I robot, where the robot is saving will smith instead of his daughter from the ground of a lake on fact survival calculations)

3. If we are aiming for an fully automatet system where the cars talk to each other communicating possible threads, sensoring the environment, additional sensors at crusial points like at stop signes or tunnel entrances can give information of possible threads(like playing children) and give the car the possability to avoid the thread by driving slower