Psychology
We need to think about the kind of humanity we want our robots to show.
We need to think about the kind of humanity we want our robots to show.
It is traditional in the annual Victoria Students’ Psychology Society Great Debate to find ways to pretend to have misunderstood the moot. As a member of the debating team arguing in support of the topic, “I, for one, welcome our robot overlords”, I professed initial confusion, claiming to have heard, “I, for one, welcome our rubber over-drawers”.
The real moot of the debate is a riff on a line attributed to the rather awful movie based on HG Wells’ short story Empire of the Ants. The line, however, doesn’t mention robots, but instead insect overlords.
I, for one, suspect that the embryos of our robot overlords are already with us. Victoria research methods lecturer Matt Hammond suggests that they are housed in our smartphones in the form of supercomputers. We take them everywhere, sharing experiences such as concerts and sunsets with them, all the while thinking we’re just using them to take photos and videos.
I’ve written before about University of Auckland associate professor Elizabeth Broadbent’s research into using robots to improve people’s lives. Examples include the fluffy seal-like robot Paro, trialled as a companion robot for patients with dementia. Recently, she’s been looking into iRobi, which not only does such things as measure the heart rate and breathing of pulmonary disease sufferers, but also offers reminders about medication and exercise. The early indications were that iRobi might not influence rates of hospitalisation, but it appears to encourage medication and exercise compliance.
It also, to my eyes, looks a little like a smaller, cuter Cyberman from Doctor Who, so perhaps that won’t allay fears about robot domination.
Who’s to say robot overlords would have it all their own way, either? Kumar Yogeeswaran and colleagues from the University of Canterbury suggest that our overlords will experience the same biases that we hold towards certain members of the human race. Their research involved a fairly standard go/no-go method in which participants were presented with pictures of people (or in this case, a mixture of people and robots) and asked to “shoot” if the target was holding a gun and refrain if the target was holding something else, such as a cellphone or soft drink. You can see the relevance of this kind of research for the high-profile debate in the US about police shootings of African Americans. Sure enough, the participants typically “shot” dark-skinned and armed human targets more quickly than armed white targets and were less likely to shoot unarmed white targets.
In the university’s research, the colour of the robots was calibrated to different racial mixes by using photos of humans of different colourings. Participants were more likely to apply to darker-coloured robots race labels such as “black” or “Asian”. But poor “robots of colour” experienced the same race-based bias we see with humans: unarmed
We may need diversity among our robots and anticipate a #MeToo reckoning if we screw it up.
“robots of colour” were more frequently targeted than “Caucasian” robots and they were more quickly taken down if they were armed, too.
One of the implications, the team suggests, is that we need to be careful about how we design our future overlords. People will invest them with value and meaning but also bias based on their personal preconceptions. We may need diversity among our robots and anticipate a #MeToo reckoning if we screw it up.
Although it’s a trope of sciencefiction books and movies that robots will rise up and try to terminate us, Hammond points out that there’s also a trope in which robots want to be more human. With any luck this will save us from the Daleks.