Build | Fix | Re-use


Powered by article titled “Why do we find it hard to torture robots?” was written by Alex Hern, for on Monday 16th March 2015 12.00 UTC

In still pictures, DAR–1 (pronounced ‘Darwin’) doesn’t look like the kind of robot that might encourage empathy.

The machine’s six spindly legs lend it an uncanny arachnoid appearance, and with no case for modesty’s sake, the exposed circuitry doesn’t hide its electrical heart.

But when the power is on, everything changes. The spiky lump of metal and plastic becomes a timid creature, locking eyes with whoever is in front of it, tracking their face and pulling back if they get too close. The obviousness of its machinery doesn’t go away, but shrinks into irrelevance.

Pull back the curtain, and the beating heart of DAR–1 is a Raspberry Pi computer, connected to a camera on an articulated head unit and six legs. It runs an open-source computer vision package to scan the area in front of it for faces, and then, when it senses one in its vicinity, locks eyes with it. If the face gets too big, it will pull back, but for the most part, that’s it.

Pulling back the curtain isn’t the point, though. DAR–1 is the creation of roboticist Ray Renteria, who introduces himself to the SXSW festival in Austin, Texas, as an amateur magician. And just as the work of a magician is focused around misdirection, and a control of the context in which magic is performed, so too is DAR–1 an exercise in how simple, mechanistic effects can be imbued with life with just the right presentation.

The work starts in Renteria’s blurb for DAR–1, a part of the festival’s “Robot Petting Zoo”. The Raspberry Pi and laser-cut legs (both, incidentally, produced in England, leading Renteria to describe the machine as a “British invader”) aren’t mentioned. Instead, visitors are primed to treat the robot as a fellow living being from the off.

“Curious about people, he’ll study your eyes and your smile with the intensity of a focused child,” it reads. “He’s shy, though. If you get a little too close to him, he’ll get nervous and try to back away. See how long you get him to keep following your eyes by looking deep into his.” Similarly, if you ask Renteria why the robot has a permanent shiver to its movements, there’s a technical answer – a particular variable hovering between fully-on and fully-off leads to motors being rapidly engaged then disengaged – but also an anthropomorphised one: “he’s nervous”.

At the same time, says Renteria, “he’s a robot, he’s proud of being a robot, so you’re not going to talk to him, you’re not going to call to him to try to get his attention.”

As well as the petting zoo, DAR–1 made an appearance at another SXSW event with a slightly less cuddly name: a talk run by Future reporter Richard Fisher titled “Would You Torture A Robot?” Helpfully, for those planning to put theory into practice, the petting zoo opened just half an hour after the torture panel closed, in the same building.

Fisher explained that, despite the fact that we all know robots can’t feel pain, there’s an increasing amount of experimental evidence which shows that people empathise with them anyway, often to the extent of being irrationally opposed to ‘hurting’ the machines.

He cites a workshop run by MIT researcher Kate Darling, who asked people to play with Pleo, a robot designed to mimic a baby dinosaur. “Yet after an hour allowing people to tickle and cuddle these loveable dinosaurs, Darling turned executioner,” Fisher writes. “She gave the participants knives, hatchets and other weapons, and ordered them to torture and dismember their toys.

“The people refused to hurt their Pleo with the weapons they had been given. So then Darling started playing mind games, telling them they could save their own dinosaur by killing somebody else’s. Even then, they wouldn’t do it.”

Robots reveal some uncomfortable truths about the way we feel empathy. Little tricks in the coding of Pleo and DAR–1 tickle just the right part of our brain to make us think that they deserve our protection, while far more advanced robots do nothing. When was the last time someone felt ethical concern about damaging a production line drone?

But even the most functional robots can still become something that people feel strongly about. A report by researcher Julie Carpenter in 2013 revealed that the attachment soldiers felt to bomb disposal robots was having very real effects.

“Soldiers told her that attachment to their robots didn’t affect their performance, yet acknowledged they felt a range of emotions such as frustration, anger and even sadness when their field robot was destroyed.

“That makes Carpenter wonder whether outcomes on the battlefield could potentially be compromised by human-robot attachment, or the feeling of self-extension into the robot described by some operators. She hopes the military looks at these issues when designing the next generation of field robots.”

Perhaps the answer is to do the opposite, and embrace humanity’s ability to develop an emotional connection with anything. After all, what’s the real difference between loving a robot and feeling sad when your iPhone breaks? © Guardian News & Media Limited 2010

Published via the Guardian News Feed plugin for WordPress.

Leave a Reply