Will gender biases reduce, increase, or will we witness much of the same?
According to some, AI replicates existing biases because it lives off the data we feed it
Fei-Fei Li, professor of computer science at Stanford University, claims Artificial Intelligence (AI) and the “fourth industrial revolution" will have an impact on every aspect of life in the not-so-distant future. The thought of living in a world dominated by AI and machine learning is both exciting and terrifying. While this transformational technology will certainly affect the functioning of consumer markets and the workforce, we ask what it might do for gender equality. Will gender biases reduce, increase, or will we witness much of the same?
According to some, AI replicates existing biases because it lives off the data we feed it. For example, the AI of a firm that historically hired male candidates immediately started rejecting female candidates, as they didn’t fit the mould of past successful applicants. According to a 2015 article in The Guardian, research showed that female job seekers were less likely than men to be shown Google ads for highly paid jobs. There’s already enough being written about the servile and obedient nature of female voice assistants and robots being developed by most technology companies—and how this reflects the needs of dominant, sometimes even perverse, men. In fact, interestingly, in the fields of law and finance, AI assistance is coded as male, to give the voice more authority and credibility. These internalized biases are not just gendered, but also have to do with race and sexuality. Is the digital world going to perpetuate centuries of prejudice, or can we correct this?
In our discussion with 100 urban millennial women, the group was divided on the subject. Straight off the bat, 45% believed it would reduce gender biases. And, exactly the same percentage believed that existing biases would stay embedded in the digital world, thereby maintaining a status quo. Interestingly, only 5% believed that technology could “enhance" gender biases. The remaining 5% were uncertain. We decided to dig deeper.
For those who believe the virtual world will make things better from a gender perspective, there were three main contributors. First, that the virtual world is quick and scalable. So, progressive thought that removes a bias could be amplified through technology and make changes bigger, faster, better. A small group of trendsetters could hope for a much greater impact than previously. Take the example of Google Maps where women are giving instructions, and that, too, on directions! Second, technology allows for anonymity and reach. Women’s voices can, therefore, be heard even if they are not physically represented in spaces, one of the biggest barriers so far. Think #MeToo.
And finally, the impact on safety. The introduction of tracking devices and other tech-abled solutions for increased security can be a game changer for women at the workplace. It will reduce the discrimination against women on the pretext of safety; and in the home space, it will give young women the opportunity to convince anxious family members that they are “visible" and safe. A millennial respondent said: “I can push for assignments that involve fieldwork, travel and late nights without worrying about safety. My parents can even track my Uber journey now." Safety considerations hinder the progress of many women at the workplace and technology-enabled solutions can ensure women are not considered to be a “liability" any more.
Millennial women who said that the virtual world will be as biased as the physical world—because current data points will continue to fuel the future—said so with a caveat. As a 27-year-old pointed out, “The advantages of the virtual world are immense, but women will remain disadvantaged, unless we acknowledge the biases that can creep in and correct for them. Take, for example, Sophia the robot—and how she has been built to be conventionally feminine."
She is clearly not alone. Gina Neff, associate professor at the Oxford Internet Institute, believes creators need to “design for use, plan for misuse, and prevent abuse". There is a growing opinion that increased representation and diversity in the technology sector, as well as in training data (a 2018 article on the Oxford Alumni website) is a must for the virtual world to correct for embedded gender biases.
The good news is that this is already being talked about. In a 2018 World Economic Forum article, Samir Saran and Madhulika Srikumar say that “to build an equitable world, which will be inhabited by women, men and machines, the global community needs to script norms around the fundamental purpose, principles of design and ethics of deployment of AI, today". They believe the involvement of a “multiethnic, multicultural and multi-gendered ethos" in the design and creation stage of AI models could deliver the equity we need and desire in future machines, and prevent the perpetuation of patriarchy in technology. Millennial women are on board with this.
It is only if we hear and respond to these voices in time that we can hope to create a non-patriarchal, unprejudiced virtual world—a post-gender world that millennial women expect. Else, we hate to say, 10 years out we will still be asking the same questions, and fumbling, frustratedly, for answers.
The Millennial Girl is a column based on an online survey conducted with over 100 urban, working millennial women to uncover their attitudes and opinions about the workplace.
Anuradha Das Mathur is founder and dean of the Vedica Scholars Programme for Women, and a Yale Greenberg World Fellow 2016. With inputs from Mohini Gupta.