The problem of gender stereotypes in robotics

One of my favorite topics to bring up during my lectures and classes is the issue of gender stereotypes that continue to be reproduced in the production of realistic robots.

This has been the subject of a previous post, but last week, Matt Simon published an article in Wired that revisits this discussion. In It’s time to talk about robot gender stereotypes, Simon begins his analysis by commenting on how robots are, after all, a powerful "blank slate" that we create, because we can customize them however we want. That is, we can create robots to help us in our daily lives, robots that are rude, or even robots that are identical to us, as is the case with Hiroshi Ishiguro's Geminoid line. In this sense, Simon comments that "the robot is not just a mirror of its creator, but of our entire species: what we make of machines reflects who we are."

And this means that the way we are already corrupting this process by creating robots with exaggerated and superficial gender stereotypes says a lot about how we are not necessarily creating a mirror of ourselves, but rather building a hall of mirrors that distort and exaggerate the proportions of their reflection. This is the hypothesis of Julie Carpenter, a scholar of human-machine interaction. "It’s very distorted, especially nowadays, as we are being introduced to the idea of robots, particularly the realistic humanoid robots that already exist in the world outside of science fiction."

What Simon questions is whether, in this sense, the biggest problem we face in robotics is the issue of gender. This includes the absolute prevalence of female voices for digital assistants, since research has already shown that people feel more intimidated by a male voice than a female voice when both are providing guidance (for example, in GPS navigation). And it is because of this research that scientists make the decision to materialize these biases, prioritizing efficiency over the cultural and social change they could promote by making the opposite decision.

"Robots don't have a gender—they are metal and plastic and silicone, and they are filled with zeros and ones. Gender is a complicated mix of biology, which robots don't have, and the way we feel about that biology, which are feelings robots don't have. Yet, we are already finding ways to reflect our social problems in robots. One study, for example, found that participants judged a robot programmed to perform security duties as more masculine, while they judged the same robot, only programmed to provide guidance, as more feminine (echoing gender preferences in digital assistants). The danger is that robot creators, consciously or not, may exploit gender stereotypes in an attempt to make their machines more effective—making a receptionist robot seem more feminine and, therefore, more 'welcoming,' or a security robot that has broad shoulders and, therefore, seems more 'authoritative.'"

But as Simon argues, it doesn't have to be this way. In reality, robots could very well be used to confront and initiate a change in these stereotypes. For Carpenter, "it would be great if, in some way, we could use robots as a tool to better understand ourselves, and perhaps even influence ourselves in a positive way. Globally, the social movement we are heading toward is one of equality. So why go backward? Why refer to gender norms from the 1960s?"

In reality, this tendency to assign gender to products is a problem already inherent to current capitalism and known through the concept of the pink tax, that is, the extra amount a woman pays for a product identical to a male one, but which, upon receiving a stereotypically feminine design (for example, being offered in pink), also ends up leading to a price increase. This is the case with Bic pens, for example, or when famous facial tissues with cute flower designs and pastel colors get a "masculine" touch so that consumers don't feel embarrassed by the "effeminate" characterization.

But in the case of robots, this is much more subtle than simply a choice of colors or patterns. A recent study revealed that when people were shown photographs of humanoid robots, they all consistently chose a particular pronoun to address the machines: "they referred to a robot with a straight torso using a masculine pronoun almost 90% of the time, but robots with a more pronounced waist were seen as more feminine. Broad shoulders, on the other hand, were classified as more masculine."

It is in this sense that Simon warns when thinking, for example, about a robot to be used in a department store. A simple design feature, such as a thinner waist, can already reinforce a tendency to imagine that it is a female or effeminate robot and, consequently, reproduce gender stereotypes in the workplace. "People are afraid to approach if the arms are too big or if the shoulders are too broad. They might even be afraid of the robot," argues Gabriele Trovato, a roboticist at Waseda University in Japan.

"Of course, designers won't always intentionally exploit gender stereotypes. A designer might, for example, want their home robot to look robust so that it gives the feeling that it is strong enough to lift you out of bed. It's about inspiring confidence. But size is not necessarily proportional to strength—electric motors don't grow like muscles when you exercise them. By dissociating the size of robots from their strength, their creators have an opportunity to reinforce the inhumanity of these machines, physiologically speaking."

Although it seems like a good solution, Simon reinforces that this may not be enough to end gender stereotypes or the tendency to assign a gender to these machines, even when they don't even look like humans or animals. And this happens even among their creators, such as one who created a small tractor-robot and treated it as "he," even though the machine is completely devoid of gender.

On the other hand, "for many robot creators, this type of tendency to assign genders makes it easier to develop a machine with a personality, and therefore it is easier for the consumer to connect with it. (That is, to buy it)," argues Simon. In the English language, by treating a robot as "it," therefore a gender-neutral pronoun commonly used for things and animals, we develop a language that is much less interesting and attractive than when we treat the robot as he or she. This is why, in the history of robotics, roboticists have always been in this conflict, and that is why it is not difficult to find examples that only reinforce these stereotypes. "It also remains our duty to demand more than these exhausted stereotypes," the author concludes.

For Carpenter, as consumers, we have the power to do better and to overcome these issues that have haunted humanity for a long time—for too long, even. So the movement ends up being a two-way street: both roboticists must start considering these elements more when developing their machines, and consumers must no longer accept this reproduction of stereotypes.

For now, however, what we see are companies in the erotic market including artificial intelligence in female humanoid robots that can have their faces changed and personalities chosen just as easily through a mobile app. In other words, we still have a long way to go.

‍