Wednesday 28 January 2015

Not just a pretty robot: gendered robots and “female” personal assistants

Corpo Automi Robot Corpo Automi Robot Bruno Cordioli, CC licence http://www.flickr.com/photos/br1dotcom/4038313726

“Not just a pretty robot, she’s artificial intelligence,” the Kickstarter for Robotbase’s “Personal Robot” boasts. The bot advertised is certainly a “she.” The screen front of Personal Robot displays various avatars throughout the video, one of which has a face like the princess emoji and all of which could stand in for the caring but feisty mother in a Pixar film. 

From this and other high profile “fembots” one might assume that all robot personal assistants are women. But they’re not all female; more than half are actually presented as male. More than one such humanoid bot is even made to be a little boy. It is unsurprising that a sex bot would be presented as male or female, but these are bots meant for essentially secretarial work. Why exactly are all these robots gendered in the first place? 

Back in the summer of 2014, another personal assistant robot, JIBO, raised two million dollars of the project’s $100,000 campaign goal on IndieGoGo. JIBO, it should be said, looks much strikingly like a lamp. Picture, as BetaBoston wrote, “A bit like a countertop cousin of Eve from “Wall-E.” But JIBO, the website for the bot makes clear, has a gender. JIBO has neither a humanoid appearance nor does he perform any that require human body parts like “eerily realistic” the Japanese sign language robot Aiko Chihara. So why is JIBO male?

Personal assistant robots are an interesting case because they are designed primarily for interactions with humans, as opposed to those built for mechanical labor. According to the Computers as Social Actors theory (CASA), popularized by the late Stanford professor Clifford Nass, humans will interact with robots and other technology in a way that mirrors social relations between humans.

In terms of the socialization of the personal assistant role, it is of course, strikingly female. According to the 2013 data from the UK Bureau of Labour, 76.6 percent of those working in the “administrative and secretarial occupations” is female.

What about the secretaries’ robot counterparts? Not the same breakdown. Surveying eight personal robots in currently production or preproduction, all but one was explicitly gendered but five out of the eight were male. Two, the Roboy and Nao were not only male but were marketed as boy-like.

There’s Luna, the personal robot with a pear-shaped body and expressive eyebrows, referred to as a “she” throughout the creator company Robodynamics’ Kickstarter ad. (You would not be remiss in noting a crowd-funding trend in the production of these personal bots.)

Pepper is a humanoid “conversational bot” with large blue eyes, meant for greeting customers and who, “at the risk of disappointing you, he doesn't clean, doesn't cook and doesn't have super powers.” As in, don’t expect this “male” Pepper to be a regular Rosie the Robotmaid out of the Jetsons.

Users of the gender-neutral Roomba, the automated vacuum cleaner, often assign their robot cleaner a gender even while the product creators do not. As a complete aside, there are entire forums on what names and genders people choose for their Roombas; reading through them is highly recommend as a way to spend an hour. This relation to the Roomba implies that people will impose a gender on a technology that is domesticated as a part of their daily lives even if the product is not marketed as gendered.

Nothing about a design of a robot requires gender, of course, except for, perhaps, us. One of the few non-gendered personal assistant bots formerly on the market is the Karotz, shaped like a cartoon rabbit. Launched in 2005, Karotz will be discontinued starting February 2015, as the company who created it wants to focus more on “humanoid robotics.”

Numerous studies show that people perceive “male” robots differently than they do “female” bots. As Motherboard put it in fall of 2014, “We’re sexist towards robots.” In a study from researchers Benedict Tay, Younbo Jung and Taezoon Park from Singapore’s Nanyang Technological University and Soongsil University in South Korea found that participants rated robots more highly if they were doing tasks in line with their gender identity. Robots with a “male” appearance were perceived as better suited for stereotypically male tasks such as security work while “female” robots were rated higher for stereotypically female roles such as caretaking.

According to the same study, the perceived personality of the robot (think: extroverted versus introverted) also mediated participants’ impressions of the bot. So we are not just sexist towards robots – we judge robots on many of the same threads we use to judge other humans, as the CASA theory predicts.

Participants not only judged robots work effectiveness differently depending on gender, but also anthropomorphized a personality in line with the perceived gender. In a study by two researchers at the German University of Bielefeld, robots with a “male” appearance were perceived as having stereotypically male traits and the same was true of the “female” bots. The researchers Friederike Eyssel and Frank Hegel did not explicitly tell the participants the gender of the robot, but varied the length of the robot’s hair in order to indicate one gender or another.

Another study at University of Bielefeld by Dieta Kuchenbrandt and others supported this finding that human-robot interactions (HRI) depended on how the human perceived the robot’s gender. In this project, participants completed a typically male or female task under instruction from a robot. The researchers displayed the gender of the robot only through its voice and name; the bodies of the male and female robots were identical.

Yet as with other humans, we stereotype robots on far more than gender. This may explain the number of boy robots, such as Nao the “daily companion bot,” built to be ingratiated into daily family life. A 2006 study by researchers Aaron Power and Sara Kiesler at Carnegie Mellon found that participants perceived humanoid robots with a baby-face as more socialable but less competent than more “mature” looking robots. Therefore, a childlike robot might be best suited for hanging out at home but not perceived as helpful as an assistant.

In all of the above studies, aspects of the robots humanoid appearance changed the way humans reacted to them. This is not what feminists, optimistic about the future of robots, predicted.

“The cyborg is a creature in a post-gender world,” wrote Donna Haraway in her 1985 Cyborg Manifesto, “I would rather be a cyborg than a goddess.” Haraway was speaking in part to the identity politics of contemporary feminism – driving home the point that there should be no one singular female ideal but rather an amalgam. (Let’s note here that as well that cyborgs are technology-enhanced humans, not synonyms for robots.)
Yet Haraway’s message was taken by some “cyborgfeminists” such as Sadie Plant, that machination would liberate women. As Plant said to Wired in 1997, “There's a long-standing relationship between information technology and women's liberation.”

The personal robot advertised today is no creature of post-gender world, as Personal Robot and JIBO illustrate. The robots are instead products of our existing social interactions, mirroring our biases right back at us.

Eve Ahearn is a 2015 MSc candidate at the Oxford Internet Institute.

www.eveahearn.com

Twitter, Facebook
Terms & Conditions, Privacy, Cookies

x