Facial recognition technology and identity politics
17 August 2019
Recently published news reports and legal studies have revealed that the US government has been violating basic democratic rights by using facial recognition to monitor and track the public. These reports have also shown that facial recognition is used widely as a preferred form of police biometrics, i.e., the science of identification and tracking through facial signatures and other kinds of unique individual measurements such as finger, palm and voice prints and dental and DNA profiles.
Every level of law enforcement—from city and state police departments to federal border patrol and military-intelligence—has been participating in the mass collection and analysis of facial images. This includes photos taken for government-issued ID cards and driver’s licenses and others captured from hidden surveillance cameras in public places such as border crossings, highways, parks, sporting events, sidewalks and airports as well as those scraped from social media accounts.
Georgetown Law published a study on May 19 called “America Under Watch” which documented the widespread use of facial recognition by the city governments and police departments of Detroit and Chicago. As part of its conclusion, the Georgetown report said, “real-time video surveillance threatens to create a world where, once you set foot outside, the government can track your every move. For the 3.3 million Americans residing in Detroit and Chicago, this may already be a reality.”
However, just as public outrage over these revelations had begun to emerge and several cities have been forced to ban the use of facial recognition—including San Francisco and Oakland, California as well as Somerville, Massachusetts—it became clear that the latest media exposures have a twofold political purpose.
On the one hand, the most conscious sections of the ruling establishment are concerned about the explosive reaction of millions of people to unfettered 24/7 state monitoring of their activities and whereabouts. Therefore, significant political pressure is being applied to force Congress to adopt as soon as possible a federal regulatory framework for the expanding biometric surveillance apparatus.
On the other hand, Democratic Party representatives and their supporters are using identity politics to conceal the serious implications of secret facial profiling and the trend toward a police state that it represents. By focusing exclusively on studies indicating race and gender bias in facial recognition tools, the Democrats are seeking to divert public anger away from a mass struggle by the working class in defense of democratic rights and into support for laws that will legalize the surveillance.
Among the most often referenced of the race and gender bias studies has been the MIT Media Lab Gender Shades project by Joy Buolamwini—a self-proclaimed “poet of code” and campaigner against “algorithmic bias”—who published her first results in 2017. The initial Gender Shades analysis of facial analysis technology from IBM, Microsoft and Face++ showed that “male subjects were more accurately classified than female subjects,” “lighter subjects were more accurately classified than darker individuals” and “all classifiers performed worst on darker female subjects.”
A follow-up study in 2018 by Buolamwini also showed that facial recognition tools from Amazon and Kairos exhibited the same trends performing “better on male faces than female faces” and “better on lighter faces than darker faces” and “have the current worst performance for the darker female sub-group.”
A similar—although far less scientific—test was performed by the ACLU on Amazon’s Rekognition facial analysis software in July 2018. A database of 25,000 mugshots was compared against public photos of every member of the US House and Senate. Amazon’s tool returned 28 false matches identifying them “as other people who have been arrested for a crime.”
The ACLU report then said, “The false matches were disproportionately of people of color … 40 percent of Rekognition’s false matches in our test were of people of color, even though they make up only 20 percent of Congress.” Since Amazon’s Rekognition is currently being used as the facial analysis tool of choice by city, county and state police departments across the country, false identification rates of minorities are of legitimate concern among workers and young people.
However, the political objective behind the Gender Shades and ACLU studies is not to prove that facial recognition should be stopped immediately, but to insist that the technology can be improved upon. The ACLU says that what is needed is “transparency and accountability in artificial intelligence” and a moratorium on law enforcement use of facial recognition until “all necessary steps are taken to prevent them from harming vulnerable communities.”
In other words, according to the ACLU, police surveillance with facial recognition of the working class is fine and can go forward with appropriate technical adjustments that reduce false identification of minorities and with the adoption of acceptable federal government guidelines for its use.
Essentially, the MIT and ACLU findings have been seized upon by the New York Times, the Democrats and organizations such as the Democratic Socialists of America as a means of burying fundamental democratic questions beneath race and gender politics. What they are advocating is, in essence, a racialist ideology and politics that serve to divide the working class and prevent a unified struggle against the growing threat of a police state in America.
For example, in entirely predictable fashion, the Times published a snide article on February 8, 2018 with the headline “Facial Recognition is Accurate, if You’re a White Guy,” which featured Joy Buolamwini and her Algorithmic Justice League.
What becomes clear in the course of the Times interview with Buolamwini is that she is on a mission for “inclusion” of minorities and women into the corrupt corporate world of government-sponsored artificial intelligence surveillance technology. Through the efforts of the Gender Shades study and others, organizations like IBM and the Ford Foundation—who talk to the Times about how they are “deeply committed” to “unbiased” and “transparent” spying on the public—get to pose as “progressive” companies.
On July 10 of this year, the Times also published a comment by its graphics editor Sahil Chinoy headlined, “The Racist History Behind Facial Recognition,” that mechanically and ahistorically sought to draw a direct line of causality between the reactionary eugenics and racialist pseudoscience of the 19th and early 20th centuries—that claimed head shape and facial morphology characteristics were predictive of behavior and mental capacities—with modern biometrics.
The purpose of the Times reporting is to show that, through the anti-bias efforts of Buolamwini and others who call for “fairness and inclusion” in facial recognition, the corporate partners of the FBI and the surveillance state itself can be convinced of the value of “standards for accountability and transparency.”
However, as was clearly shown in the second Gender Shades study, the result of this campaign against race and gender bias has been the improvement of software algorithms by the developers. Although this aspect of Buolamwini’s study has been given little attention, the 2018 results showed, “Within 7 months, all targeted corporations were able to significantly reduce error gaps … revealing that if prioritized, the disparities in performance between intersectional subgroups can be addressed and minimized in a reasonable amount of time.”
This was also echoed in a statement sent to Buolamwini by IBM which, according to the Times, said that within a month of her second study results, the company “will roll out an improved service with a nearly 10-fold increase in accuracy on darker-skinned women.”
The race and gender bias campaign has also fueled efforts by Democratic Party politicians across the country to cover up their own role in the creation and maintenance of the surveillance state apparatus by posturing as opponents of the false identification flaws of the software.
The racialist character of identity politics was on full display during a hearing of the House Oversight and Reform Committee on May 22, for example, when Rep. Alexandria Ocasio-Cortez (Democrat of New York)—a member of the Democratic Socialists of America (DSA)—concluded her questioning of the panelists with a stage-managed exchange with Buolamwini who was a key witness. After being asked by Ocasio-Cortez what demographic facial recognition tools are “mostly effective on” and “who are the primary engineers and designers of these algorithms?” Buolamwini responded multiple times with “white men.”
Ultimately, the efforts of the Democrats, the DSA and identity politics advocates are aimed at diverting the anger of the working class against the mass surveillance into demands for congressional action that would make it legal to spy on the public with facial recognition tools as long as it is “fair and unbiased.”
The defense of democratic rights cannot be entrusted to these representatives of the affluent middle class who are employing methods of deception to conceal their own moneyed interests in getting on board with the private surveillance industry as well as the true meaning and historical implications of what is unfolding in society.
The growth of extreme economic inequality, the unending US wars in the Middle East and elsewhere and the evolution of the administration of President Donald Trump toward authoritarian rule—with the support of the Democratic Party—are all aspects of decaying democracy in advance of a massive confrontation between the working class and the ruling establishment in America and internationally.
It is in preparation for this conflict that the surveillance state is being erected and perfected. Workers, students and young people must unify across all national, racial, gender and language differences—independently of the government, the corporations and the middle-class pseudo-left—to make their own preparations for the major class battles now on the horizon.