Much of the debate about autonomous vehicles (AVs) has focused on the driverless dilemma – who to save, or kill, in no-win crash situations.
This subject is often explored via a thought experiment called The Trolley Problem, which imagines a runaway train and five people tied to the track. If you intervene by pulling a lever, the train will switch to a track with just one person.
Numerous studies, notably The Moral Machine, suggest broad agreement that: 1) humans should be saved over animals; 2) the lives of many should outweigh the few; and 3) the young should have precedence over the old.
However, in this article for Robotics Business Review, Julian De Frietas, of the Department of Psychology at Harvard University, and Sam Anthony, of Perceptive Automata (a company specialising in human behaviour in robotic systems), question the merit of applying such thinking to driverless cars.
“There are two problems with the trolley dilemma – first of all, it’s a distraction to the work that is being done on making AVs safer, and second, it has this built-in assumption that AVs can see the world perfectly,” says Anthony.
Initially this seems cavalier, an affront to the mainstream view that the driverless dilemma is vital to the debate. It is certainly an issue that cuts through with the public.
De Frietas goes on to assert that such dilemmas – situations where you have the time to make a considered decision as to who to kill but can’t use that time to avert it – are rare.
A better approach, he argues, is aiming to avoid harm: “That means that if most of what you’re doing on the road is just avoiding more mundane things, then optimizing to that goal will cover you.”
There’s a lot to digest there, particularly considering the infamous comments reportedly made by a Mercedes-Benz executive at the 2016 Paris Motor Show about saving the driver and passengers over pedestrians.
On a personal note, I’ve been driving for 25 years and have only found myself in something resembling a trolley dilemma once. A car pulled out in front me – pedestrians left, solid traffic right. I almost managed to stop but went into the side of the car that pulled out. We all walked away but, believe me, one trolley dilemma in a lifetime is more than enough.
In the same situation, what will a driverless car’s programming tell it to do? Will this vary across different makes and models? Should vehicle owners have any control over the settings?
Perhaps Anthony and De Frietas deserve credit for scrutinising the driverless dilemma, but their stance only reaffirms my view that it should be the touchstone for all autonomous vehicle development.