Coded Bias — an 85-minute documentary directed by Shalini Kantayya (A Drop of Life, 2007 and Catching the Sun, 2015)—began streaming globally on Netflix in April. The film premiered at the Sundance Film Festival in early 2020 and has since received positive reviews and several awards including Best International Documentary at the 2020 Calgary International Film Festival.
The central argument of Coded Bias is that the computer algorithms used in artificial intelligence, machine learning and facial recognition systems are “embedded” with the race and gender biases of society as a whole.
In developing this theme, the documentary features the work of Joy Buolamwini, an African-American MIT Media Lab researcher and founder of the Algorithmic Justice League who has specialized in uncovering such biases in facial recognition technologies.
In her research, first published in 2017, Buolamwini showed that the facial recognition algorithms developed by IBM, Microsoft, Face++ and Google were based on a limited data set and less accurate when identifying darker and female faces than in identifying lighter and male faces. As is depicted in the film, in spite of state utilization of such tools in violation of Fourth Amendment rights, Buolamwini’s research helped the developers improve the accuracy of their face recognition algorithms.
Buolamwini’s work has been heavily promoted by the New York Times and other media and political advocates of identity politics. She was a key witness at the Democratic Party-run House Oversight and Reform Committee hearing on facial recognition in May 2019. At this hearing, Rep. Alexandria Ocasio-Cortez (Democrat of New York)—a member of the Democratic Socialists of America (DSA)—asked Buolamwini what demographic facial recognition tools are “mostly effective on” and “who are the primary engineers and designers of these algorithms?” to which Buolamwini responded with “white men.”
Through interviews with Buolamwini and other activists opposed to “algorithmic discrimination,” Coded Bias paints a picture of a dystopian society in which white male bias against minorities and women are deeply rooted and the source of run-away advanced technologies that perpetuate prejudice and discrimination.
Meanwhile, when the issues of equality and democratic rights are referenced, they are presented in the film as subordinate to the discrimination experienced by various identities. The word “capitalism” gets no mention and “class” is only brought up occasionally at the end of a list of other identities that include race, gender and sexual orientation.
In an effort to lend a theoretical backbone to their conceptions, the filmmakers and interview subjects focus in on what they consider to be the origins of the demographic biases being transmitted into advanced technologies. Buolamwini says, “My own lived experiences show me that you can’t separate the social from the technical. ... When you think of AI, it’s forward looking, but AI is based on data and data is a reflection of our history, so the past dwells within our algorithms. This data is showing us the inequalities that have been here.”
This idea that race and gender bias within data “is a reflection of our history” is further developed in an interview with Meredith Broussard, data journalism professor at New York University and author of the book Artificial Unintelligence .
Broussard concentrates her criticism on the fact that AI was launched in 1956 in the math department at Dartmouth College. She says that “there were only maybe 100 people in the whole world working on artificial intelligence in that generation” and “one faction decided that intelligence could be demonstrated by the ability to play games and specifically the ability to play chess.”
Showing school yearbook photos of the Dartmouth math students who pioneered artificial intelligence—and without naming them—the message of Coded Bias is clear: the problem with AI is the fact that a group of white men initiated it. The fact that John McCarthy—the Dartmouth professor who coined the term artificial intelligence—and the others, such as Marvin Minsky, were pioneers in the field and made early contributions to the science is entirely dismissed by Broussard. Their significant and lasting discoveries in the field are not even discussed.
As she asserts, “Our ideas about technology and society that we think are normal are actually ideas that come from a very small and homogeneous group of people. But the problem is that everybody has unconscious biases and people embed their own biases into technology.”
The theory of unconscious bias is dominant among the group of researchers and specialists selected by Kantayya for inclusion in the documentary. Alongside of other reactionary racial identity conceptions such as “white privilege,” the proponents of unconscious or implicit bias theory insist that the source of discrimination and inequality in society stems not from the socio-economic and class divisions within capitalism but from the bigotry, prejudice and stereotyping that is deep-seated in all humans without their being aware of it.
Although the film makes reference to the undemocratic implications of facial recognition surveillance, as well as the invasive use by corporations of advanced algorithms for business purposes with no concern for the rights of the public, Coded Bias repeatedly returns to the themes of race and gender identity. In the end, the political proposals advanced by those being interviewed and the message of the film itself is that the use of the surveillance tools must be made legal by getting Congress to enact laws that govern their use.
Meanwhile, the most criminal implementations of advanced algorithms and artificial intelligence under capitalism—such as their use for the purpose of militarism and imperialist war—are not even mentioned. While the deployment of unmanned drones and battlefield robots represent the bleeding edge of the technologies discussed in Coded Bias, significantly the AI programs of the Pentagon and US Defense Department do not find their way into the Netflix documentary. These are thoroughly “respectable” and establishment figures, who appear to have no difficulty with American imperialist violence around the world.
As was the case with the Netflix film The Social Dilemma , this documentary about AI and algorithms allows its “experts” to make sweeping statements about technology intended to stir up the anxious upper middle class. The entirely one-sided approach of Coded Bias rejects any progressive element to the technology or the fact that its undemocratic and repressive implementation—just like the persistence of prejudice and discrimination—are rooted in the profit system.