27 Jul 2018

Robots and racism: Are robots too white?

From Jesse Mulligan, 1–4pm, 1:16 pm on 27 July 2018

A new study from Canterbury University and Monash University academics shows humans have the same racial biases towards robots as they do towards other humans.

A Ping Pong Playing Robot at Tokyo International Robot Exhibition, Nov 2009

A Ping Pong Playing Robot at Tokyo International Robot Exhibition, Nov 2009 Photo: By Humanrobo CC BY-SA 3.0

One of the study's authors, Dr Christoph Bartneck, told Afternoons’ Jesse Mulligan nearly all humanoid robots are either white or grey. 

No caption

Photo: Wiki Commons

“One of my co-authors was about to give a presentation about robotics and he was looking for images of brown robots and he couldn’t find any,” he says. 

“For all of you I encourage you to open your browser, type in ‘android’ or ‘human robot’ and what you’ll find is that all the images being shown are pretty much white, a bit of grey here and there.” 

He says they set about designing a study to test if people showed the same racial biases for robots as they did for people. 

“Asking direct questions doesn’t really work very well,” he says. 

“You can’t really directly ask somebody ‘hey are you racist? The answer would of course be ‘no I’m not a racist’.” 

They decided to replicate a previous study that had been done on racial biases, where participants only have time to react, not really think about their response. 

Robot

Robot Photo: Alex Knight via unsplash.com

“You take participants and you put them in the role of the police officer and you show them images of people and these people have either a gun in their hands or they have a benign object like a soda can. 

“The participants have to make a choice between shooting and non-shooting. 

“When the study was originally conducted in the US in 2002 it showed that when you show black people, people were actually much quicker to respond than when reacting towards white people.” 

He says it’s a reliable study that has been replicated many times.

“What we did is actually we completely replicated the study just to make sure we did our job well, that we get similar results to the original study, and then we added white and black robots to it.” 

He says they used images of the 58cm ‘NAO’ robot from SoftBank Robotics and changed nothing but the colour scheme.

78805094 - programmable humanoid robot nao on robotics expo in moscow, russia

SoftBank Robotics' NAO robot. Photo: prescott09/123RF

“These were humanoid robots, they have arms, legs, the same kind of body structure as a human. 

“We have a tendency to treat them as though they are social actors, as though they are alive. 

“And behold, the exact same biases that you observe towards images of humans also occur when you show them images of robots.” 

He says there’s no reason for all the robots to be white. 

“Robots are being developed all over the place … everywhere really, but all the developers seem to agree on that white is the appropriate colour for a robot, or maybe grey. 

“I know a lot of my colleagues who work in the field and again, they’re a very diverse bunch of people. 

“There was no particular reason why [the robots] all should be white.” 

Dr Bartneck says it’s a problem that developers should be looking to solve. 

“If you imagine in a future where we have more and more of these robots entering our lives - and our society is diverse we have different ethnicities, different races - but if all the robots are white that would certainly be sending the wrong signal, wouldn’t it. 

Teams of Nao robots compete in RoboCup, the robot soccer world cup.

More NAO robots. Photo: RNZ / Alison Ballance

“It should at least be possible to have a choice about what kind of colour your robot has, if it’s all just white that really suggests some form of white supremacy that we really do not want to have. 

robot love

Photo: 123rf

“Racism is a real problem in many countries and many people suffer from it, and we should do whatever we can to improve this. 

“Just last year we had this big discussion about the emojis which were all yellow, and there was a bit of an outcry. 

“The companies have become more sensitive to this and are offering now emojis in different kinds of colours.” 

A new study from Canterbury University and Monash University academics shows humans have the same racial biases towards robots as they do towards other humans.

One of the study's authors, Dr Christoph Bartneck, joins us to talk about the evidence, the implications, and what we should be doing about it.

Get the RNZ app

for easy access to all your favourite programmes

Subscribe to Jesse Mulligan, 1–4pm

Podcast (MP3) Oggcast (Vorbis)