If Siri sounded like a man, would you think differently about your iPhone’s built-in helper? Or what about Alexa, the voice of Amazon’s Echo technology—does it matter to you that she’s a “she”?
Neither Siri nor Alexa are women, of course, but rather computer programs designed to mimic the speech patterns of human females. But before you laugh off the questions from that first paragraph, consider this: Designers and coders and executives and unknown multitudes of other workers at Apple and Amazon settled on the idea that Siri and Alexa would be women—and they did so for specific and self-aware reasons.
That’s a fascinating thought. Here’s another: In the late 1990s, German automaker BMW sent out a recall on the voice-navigation systems for 5-series cars because some customers didn’t like taking driving directions from a woman’s voice. Sexist? You bet. But that quirky little story—along the with examples of Siri and Alexa—show us that people tend to develop all kinds of perceptions about their computers and that, eventually, those perceptions begin to inform users’ interactions with their computers.
As you might have guessed by now, one of those perceptions has to do with gender. Strange as this all might sound, that’s the subject Marek N. Posard’s article “Status processes in human-computer interactions: Does gender matter?” (available in the August 2014 issue of Computers in Human Behavior).
The short answer to Posard’s question is “yes.”
To find out, Posard designed a study based on status characteristics theory (SCT), which holds that groups of people widely share beliefs about different social groups. According to SCT, these beliefs create expectations about the members of the group, whether or not the expectations reflect reality. As an example, Posard pointed out that studies have shown that participants often rate the work of men and women differently even if their performance is the same. The differences, according to SCT, stem from these preformed beliefs held by participants.
After noting that SCT also applies to how people view their computers, Posard laid out the specifics of his study. Participants from a Mid-Atlantic university conducted a “gender-neutral” task with a partner and then rated how well the partner did. The catch was that the partner was a computer depicted as either a woman or man—Julie or James. Of the 63 participants, 46 were women and 17 were men. The participants were informed that they were testing a computer program that automatically grades exams taken by students. The instructions informed the testers that the program could “learn” from its interactions with humans. Following the test, participants were asked to share their impressions of James and Julie. Finally, participants were asked how much they thought the computer system would cost.
The results of the test showed that the participants had roughly the same level of confidence in Julie as they did in James. That’s an encouraging sign for humanity, and what’s more, the participants rated the performance of the two computers similarly. However, there was a big difference—$1,490 on average—in how much they thought each computer would cost.
So why the difference? Remember, the only difference between the computers was that one was named Julie and the other James. In both cases, testers performed gender-neutral tasks with computers that were supposedly capable of learning from past interactions with humans. And yet, on average, the participants thought that the “male” computer would cost almost $1,500 more than the “female” version. According to Posard, one reason for this could be that the participants, whether they knew it or not, referenced “broader patterns of inequality within society”—such as gender inequality and the pay gap—while they were taking the test.
What do you think? Do we, as users of technology, bring certain pre-conceived notions into our interactions with computers, much like SCT hypothesizes? Do we treat computers inequitably the way some of us do with different social groups? If so, what does that say about us? Is it a good or bad thing?
Regardless of the reasons, and regardless of how you answer those questions, Posard’s test showed that gender does matter to people in at least one way when they interact with their computers. The next time you’re using your smart phone, your computer or even your new high-tech car, you might not be able to look at “her” or “him” the same way.