Should driverless cars make life-or-death decisions?

While the benefits of self-driving vehicles are many, Brett Debritz notes that there is one significant barrier to overcome.

An driverless car is demonstrated at the CES trade show in Las Vegas. John Locher / AP

Imagine arriving at Dubai or Abu Dhabi airport, tired after a long flight, and getting to the taxi stand to discover that your cab doesn’t have a driver.

He hasn’t stepped out for a smoke; this is entirely the way it’s meant to be. You get in the car, speak a few words into a micro­phone, then sit back and relax as you are safely and quickly delivered to your destination.

It sounds like a dream scenario – a stress-free journey during which you don't have to worry about erratic driving, giving directions or having to discuss the latest cricket scores or the weather with the driver.

Well, if some of the pundits – whose opinions are backed up by big money from the likes of Google, Uber, Tesla and probably, but not officially, Apple – are right, that could be happening very soon.

Indeed, Dubai is among a growing number of cities in the world experimenting with driverless car technology, with a target of 25 per cent of all car journeys being driverless by 2030.

The UAE is a great candidate for driverless vehicles: it has a wealth of people with disposable cash and an insatiable desire for the latest gadgets, a public love affair with motor vehicles and – let’s be brutally honest – some really, really bad drivers. Our unacceptably high road toll is stark evidence of that. Surely, one would think, robotic cars couldn’t be any worse.

But there’s the rub. According to car manufacturer Toyota, and academics who specialise in the ethical choices made by machines, humans will not be prepared to accept road deaths that occur due to the judgment of a robot.

Most accidents are preventable and it is highly probable that driverless cars will be much safer than those with drivers. But even if every car on the road was driverless, some vehicles would occasionally malfunction, or something unexpected would occur – such as a pedestrian running on to the road.

In this scenario, the computer driving the car might have to make a split-second choice between hitting the person on the road in front of them, or swerving into another vehicle, potentially causing the death of its own passengers and the passengers of the other vehicle.

According to Toyota, which has invested more than $1 billion (Dh3.67bn) into driverless technology, society won’t accept road deaths that are the result of machines making life-or-death decisions – even though we now accept human misjudgment as the cause of many fatalities.

Gill Pratt, who runs the Toyota Research Institute, said at the CES electronics and technology trade show in Las Vegas last week that no manufacturer was anywhere near developing the technology that would be needed to make driverless vehicles sophisticated enough to overcome this barrier.

Researchers at the Massachusetts Institute of Technology are also working on this and similar issues. Their Moral Machine website has a "game" in which users are asked to judge how they think a driverless car ought to react in 13 scenarios.

It comes down to whether the car should risk the lives of its passengers or of other road users – but the number, ages and genders of the participants change from scenario to scenario.

The person playing the game is essentially required to judge the worth of their own friends and family against that of, among others, a dog, a young boy, a girl, a pregnant woman and an elderly man.

Few of us would be willing to say that we thought one life was more precious than another, let alone actually have to make that decision for real. So what would we think about having a computer make that distinction, and what criteria would we expect it to follow?

This is the debate that we have to have before driverless cars – and any technology that takes over roles involving moral or ethical judgments – become anything other than a novelty.

On Twitter: @debritz