Artificial intelligence (AI) is the development of computer systems that can perform in ways that would typically require a level of human intelligence as a means of aiding their human counterparts (think Siri, Alexa, GPS systems, etc.). However, the gendered nature of AI programming is much less complex— it has its roots in the same social constructions as every other institution obsessed with creating a dichotomy between man and woman. All whilst ignoring the fluidity and numerous other identities gender can encapsulate, all whilst ignoring that the masculinity or femininity of language is determined by preconceived notions of what it means to be masculine or feminine. Gender roles still weigh heavily on the ways that we interpret which spaces we can occupy and which products are worthy of our consumeristic notice. AI systems are no exception.
Studies on AI programming have shown that American consumers respond best to a feminine voice. They find it to be warmer, more conversational. In addition, when presenting a male-voiced Siri, researchers found that consumers are more likely to trust the information, crediting the findings to the fact that consumers find male voices to be more intelligent. Take into account that many of Siri’s abilities (taking notes, looking up answers, scheduling appointments, subservience, etc.) resemble those of a secretary, and therein lies another factor. The subservience of these devices— the fact that they are made of us and for us is highly resemblant of biblical rhetoric that has been interpreted to excuse the oppression of women for centuries. AI systems have come to represent the literal objectification of women as a nothing more than the slender, polished trophies we keep in our back pockets.
But, these programs aren’t the perpetrators of their own sexist inclination. The real issue, like with many STEM fields, is that there is a severe lack of diversity in the computer science field. Most recently, only 17% of computer science degrees were awarded to women, down from the peak of 37% in the mid-80s. Much like a lack of diversity in film writing rooms and studios, a lack of diversity in the field of AI creation creates a lack of perspective that shifts from that of the Matt Damons of the world. And, this need for diversity on the creation end extends far beyond just the addition of women to the field. The inclusion of people of color, differently-abled individuals, those of different sexual orientations and gender identities, etc., is essential to ensuring that AI programs are adaptable and usable for the entirety of the population their meant to aid as opposed to just the most “normalized.”
In thinking about AI and how it continues forward, it’s important to remember that the programs can only be changed by those who create them. In order to see a change in what’s being created, we have to encourage, promote and uphold safe spaces for women and minority groups to pursue studies in STEM subjects. We must also be mindful of how masculine language in job postings can be used to deter women from applying. AI is supposed to be un-gendered. Let’s try to make it that way.
Featured image here.