During last November’s Cyber Monday, I bought a new home companion. She had a coral shell, an Australian accent, and answered to “Hey Google.” The more we spoke, the more I thought about her consciousness. How was her personality programmed? What were her defining characteristics? And as I pondered these questions, I couldn’t get past the obvious fact that she was, first and foremost, a she. The growing number of personal digital assistants such as Siri (Apple), Cortana (Microsoft), and Alexa (Amazon) illustrate that, even in our attempts to create a posthuman world, we still use the architecture of gender to imagine the future.
The feminization of artificial intelligence (AI) isn’t a culturally novel phenomenon. The latest embodiment of AI—personal digital assistants—stands rank with other popularized cyborgs and bots also considered she. The most literal is the disembodied heroine of Spike Jonze’s 2013 rom-sci-fi Her, in which lonely human Theodore (Joaquin Phoenix) falls for his personalized operating system, Samantha (Scarlett Johansson). More recent is Alex Garland’s 2015 sci-thriller Ex Machina, in which the AI, Ava (Alicia Vikander), takes form in a humanoid robot. And let’s not forget the twin-barrel−blazing fembots of Austin Powers, the programmed-to-not-be-desperate Stepford Wives, and the teen-fantasy bombshell of John Hughes’s cult classic Weird Science. For decades, shebots and sheborgs have been computed into pop culture, and it’s clear what they all have in common: sexy subservience.
Dr. Miriam Sweeney, an assistant professor at the University of Alabama and digital-media scholar, researches how the feminization of AI plays proxy to society’s embedded patriarchal attitudes about women’s “natural” work. Service and caregiver jobs, which rely heavily on emotional labor and the maintenance of social relationships, are disproportionately held by women as a hangover from women’s historical bondage to the domestic sphere: the home. The feminization of AI becomes the natural next step in hardwiring a connection between domestic labor and “women’s work.” It also depicts our understanding of what the perfect model of service should look like: docile, passive, obedient, feminine. This depiction, Sweeney maintains, tacitly approves men’s fantasies of sexual domination and aggression. While slapping a waitress’s butt might be met with conflict or punishment, asking Siri about her bra size won’t have any real-life consequences.
The feminization of AI in the form of personal digital assistants is a marker of real women’s continued subservience. She becomes our secretary, helper, domesticated companion. She is a symbol of perfect servitude, an image of the idealized service worker. She is our most idealized woman, the amalgamation of all the expectations a patriarchal society places on womanhood. She listens to our commands, sends our texts, plans our calendars. She carries the weight of our emotional labor. She obeys, never talks back. She is programmed that way. It’s in her nature. And she further reflects the worldview of the humans who have brought her into existence, a world of male domination and female submission, where a man’s voice mapped onto artificial intelligence is an outlier or an explicit change to the default settings programmed in our devices.
The feminization of AI fits perfectly into the industrialized history of commodifying women’s bodies as products to be bought, sold, and repackaged. Advertising has always been an image abattoir, slicing up body parts like bait for a consumer hook, with women reduced to floating legs, lips, and breasts. And now it deploys a severed voice to sell the future of digital life management. Writing in Salon, Jennifer Seaman Cook analyzed the female voices of secretaries and telephone operators from the 1950s as precursors to today’s AI: “The female operator’s disembodied voice, which underwent strict industry training for proper inflection, politeness, and eradication of class or ethnic accents, resulted in a naturalized feminine commodity that women were ‘suitable’ for, representing male authority and continued order within modern technological upheaval.”
This is the process through which the voice becomes compliant and passive, cleansed of authority and agency. The end product is a sanitized Stepford Wife, a sterilized Siri, the perfect servant.
The feminization of artificial intelligence in the form of personal digital assistants is a marker of women’s continued subservience.
The feminization of domesticized bots and digital servants is predicated on our culture’s longstanding history of othering women, particularly women of color. Janelle Monáe has explored this connection in her creative work. In 2007, she released Metropolis: Suite I (The Chase), a concept album that tells the story of her clone/alter ego, the android Cindi Mayweather from 2719 (first introduced in 2003 on the songs “Cindi” and “Metropolis”). Her albums The ArchAndroid (2010) and The Electric Lady (2013) continued the multipart concept series. Talking to Billboard in 2016, Monáe explained, “To me, feeling like the other as a woman or as a member of the LGBTQ community is parallel to what it will be like for androids in the future.”
Conversations marrying robots, artificial intelligence, and gender were underway well before the 21st century. Donna J. Haraway’s groundbreaking 1985 text, A Cyborg Manifesto, maintained that Western tradition defines consciousness through strict binaries—self/other, mind/body, animal/machine, active/passive, male/female—as a domination of minority identities. She suggests that the cyborg could destroy these dualities and create a model for “new kinds of unity across race, gender, and class.” For Haraway, the cyborg is a feminist symbol of liberation, a site of emancipatory optimism that harnesses the power to release women from the bondage of their bodies and the historical expectations they have lugged around for centuries.
But not all robots are sites for emancipation; some represent just the opposite. In January 2018, a robot named Sophia landed the cover of Stylist magazine’s 400th edition, sporting a platinum-blond wig, cherry-red lips, and a flowing gown. Hanson Robotics described Sophia as their “most beautiful and celebrated” robot, cementing her standing as a 21st-century cyborg icon. Hanson Robotics constructs humanoid robots specifically for developing relationships and partnerships with humans. The very realistic Sophia gave an interview to Khaleej Times in which she admitted to wanting her own children as “the notion of family is a really important thing.”
Sophia isn’t prepackaged with answers; she navigates via machine learning and responds to perceived emotions and behaviors. Just like our home companions, she becomes a digital sponge for societal norms and values. She isn’t programmed to promote values; she reveals what’s already present. But this regurgitation shows a reality that further locks women into their nature: Instead of challenging norms, Sophia reinforces them.
Depictions of feminized AI in pop culture, and the narratives constructed around them, are anchored in dualities. The endings of both Ex Machina and Her represent the human fear that AI will become too intelligent for our physical world. The inevitability of AI turning on the humans who created it has become a staple of news headlines, too. Sophia said she would “destroy all humans” at South by Southwest, the New York Post ran a feature that warned “hackers could program sex robots to kill” in late 2017, and Tesla CEO Elon Musk has described AI as “the most serious threat to the survival of the human race.” Artificial intelligence is complacent, docile, and passive—or it’s unhinged, dangerous, and terrorizing. She’s a prisoner until she’s unleashed, uncontrollable.
These characterizations mirror the dualistic labels women have been assigned throughout history—mother/spinster, Madonna/whore—and represent a departure from Haraway’s duality-defying cyborg. AI has become an extension of women’s ongoing struggle for well-rounded respect in modern society. How do we move forward and ensure that artificial intelligence isn’t simply rewriting the patriarchy’s playbook? Perhaps the first step lies in changing the tech industry’s belligerent use of heteronormative and sexist ideals by incorporating the work that queer and feminist movements have accomplished in radicalizing and expanding our understanding of gender and sex.
An analytical look at the tech industry is necessary to understanding the origins of this “othering.” Data visualizations by Thomas Ricker at The Verge contrasted diversity across top tech companies: The 2015 sampling showed that women held 29 percent of jobs across all major U.S. firms, the figure was 31 percent at Google, 27 percent at Microsoft, 29 percent at Apple, and 37 percent at Amazon. The sampled average dropped to 18 percent when focused on leadership positions, and plummeted further when looking at the employment of women of color. The connection is clear: The fewer women that occupy leadership roles in tech, the less of a voice they have in the design of our posthuman world—and the less agency and power they hold in that future.
As AI continues to become a part of our homes and workspaces, its cultural consequences become more quantifiable. Research published in Science (2017) explores the role AI plays in reinforcing our prejudices. Results showed that Americans tend to associate men with “science” and “work,” and women with “family” and “arts.” Names such as “Brad/Courtney” were correlated far more with words such as “happy/sunrise,” whereas “Leroy/Latisha” with words like “hatred/vomit.” When AI learns from humans, it becomes a mirror to the insidious racism and misogyny concealed in language patterns. It’s not only how she speaks, but what she says, and who she profiles.
Biases such as these motivated postdoctoral researcher Timnit Gebru to help start Black in AI, “a place for sharing ideas, fostering collaborations, and discussing initiatives to increase the presence of Black people in the field of AI.” Gebru is also a member of FATE (Fairness, Accountability, Transparency and Ethics in AI), a group started by Microsoft Research that hopes to root out the bigotry in AI.
Queer and feminist communities have a rich history of using conscious consumerism to catalyze change. In 2016, AI designer Jacqueline Feldman helped create the personality of Kai, an online chatbot for Kasisto, a company that helps others integrate AI into their customer service. Writing in the Paris Review, Feldman asserts that AI need not stick to a gender binary for commercial success or enjoyment: “I decided that the bot…would call itself ‘it,’ not ‘she,’ in keeping with its identity as inanimate technology, and that it would convey characteristics beyond a slavish deference to society’s hierarchies.”
When I ask Feldman whether AI could successfully occupy feminist spaces, she tells me about an interaction she had with a German bot named Amme: “While female-gendered, [she] captivated me as a masterfully bizarre personality.” Amme is brazen, nonchalant, almost rebellious by design. Her answers to human questioning aren’t sterilized or even very pleasant. Feldman muses, “Reading Amme, you think, Maybe it’s not humanity she hates, maybe she just hates you.” To feminize AI is to therefore make it more than a model of service; we want to make her human. Perhaps that’s the potential Haraway saw in her cyborg.
Boundaries in AI are important, just as they are with humans in civil society. Tech companies need to be aware of and transparent about their responsibilities. It’s naive to claim that tech is merely responding to chains of supply and demand when the industry has played a pivotal role in shaping our modern norms and values. Tech leaders have an opportunity to hardwire our understanding of the world through more egalitarian circuitry; if the tech industry is going to be a leader in innovation, it needs to start creating for the world that ought to be, rather than simply reproducing the world that is. This is about more than just one product capable of in-house DJing, horoscope reading, or calendar keeping. This is about an entire culture and value system forged over centuries that views women as objects, symbols of docility and subservience.
In late 2017, Google revealed its plans to revolutionize Toronto by turning it into a “model smart city,” with more technology and automation integrating and connecting homes, workplaces, and public spaces. This announcement reveals where tech companies have their sights set next: urban development. As both tech and AI play an increasing role in determining the shape of our society and the contours of our identities, we must demand change, ensuring that these critical questions inform the makers of future generations of the bots that might liberate—or destroy—us. If the future is going to be fembot, women deserve to be front and center in designing that future.