Jason Bent and Ifeoma Ajunwa have authored recent papers I like a lot as they help to uncover and prescribe some solutions to the potential racist treatment of workers through technology as we advance into 2021. Their suggestions on how to address this form of employment discrimination come at a crucial time for workers of color. The nature of racial discord in our society reached a crescendo in 2020 and raised many questions for workers of color. The Covid-19 pandemic placed unusual health and economic burdens on black and brown workers as the insidious nature of the virus afflicted communities of color more harshly. So-called essential workers, many of whom are vulnerable people of color, were forced to risk exposure to the virus in order to perform their work duties in-person as most other workers scurried off to their homes to perform their work duties in a virtual manner. Meanwhile, militia and white supremacist groups have taken a more active role in our society as a response to the national and international protests calling for racial justice after the senseless killing of George Floyd by a police officer in Minnesota.
With the racial consequences from Covid-19 and the George Floyd protests still looming, the country will attempt to recover from the events of 2020. As these recovery efforts proceed, we must not forget that workers of color also face another racial problem, the effects from increasing technological advances aimed at giving employers greater opportunities to capitalize on the use of big data. Both Bent and Ajunwa have authored papers that examine similar concerns related to racial problems caused by technological developments as employers attempt to use algorithms aimed at achieving greater operating efficiencies. Although their suggested resolutions to this problem offer different approaches, both authors, as discussed below, give their readers an interesting take on how workers of color may be subjected by their employers to racism through algorithms and how that form of workplace discrimination should be addressed.
As we advance into 2021 with the hope of stalling the virus and reclaiming or reinvesting in many of the advances that can expand the economy and restore any business growth delayed and stifled by the pandemic, Bent and Ajunwa keep us focused by providing insights about the use of technology that can result in discrimination against workers based on race. Their contributions add to some of the recent literature advancing the concerns about the broad racial implications arising within the growth of algorithms. In the midst of the Black Lives Matter movement and other indicators that highlight how pervasive discrimination exists in our society, the use of racist algorithms has started to generate some concern. Fallible humans operating in a society involving systemic discrimination provide the input for the algorithms. Then those inputs infected with the existing racism of the past and the present can lead to racist outputs even if unintended by those using an algorithm they perceive to be a neutral application.
With general concerns for workers of color subjected to racist algorithms and following the work of technology scholars suggesting that the best way to respond to this form of discrimination is to take affirmative action steps to identify and prevent any discriminatory results, Bent and Ajunwa recommend unique solutions to address this predicament. A few workplace law scholars have started to address the discriminatory problems presented by using emerging technologies to fortify racist treatment of employees. The recent papers of Bent and Ajunwa add to those prior contributions while also offering important considerations for workers of color at such a crucial time for employers to be aware of any actions they take that might create racial concerns for their employees.
On page 824, Bent asks: “Are race-aware algorithm fairness solutions permissible under U.S. anti-discrimination law?” By posing this question, Bent begins an inquiry as to whether employers can correct for racist effects from a neutrally-applied algorithm without violating employment discrimination laws when the correction operates as a form of affirmative action. Bent navigates Supreme Court precedent finding that corrections to or failures to certify test results to prevent disparate impact liability can operate as a form of disparate treatment discrimination against the workers who would have benefitted from the certified results.
In a thoughtful and thorough analysis of the Supreme Court’s jurisprudence on Title VII disparate impact theory and affirmative action law, as well as constitutional equal protection analysis based on race for public sector employers, Bent constructs a possible framework whereby employers may escape disparate treatment discrimination liability when pursuing race-conscious changes to algorithms. Bent turns to the Court’s affirmative action law analysis to distinguish the Court’s disparate impact law jurisprudence as a means for an employer to escape liability when correcting a racist algorithm. According to Bent, an employer’s race-conscious attempt to prevent racial disparities resulting from application of an algorithm could be viewed as part of an affirmative action plan instead of an attempt to correct a disparate impact result.
From Bent’s review of the cases, an ex ante consideration of actions intended to prevent racial disparities will find better traction in complying with the Supreme Court analysis as a proper and legal action under affirmative action law. (Pp. 836-37.) Under Bent’s reasoning, the Supreme Court’s disparate treatment concerns for those workers who would have achieved certain measures under the status quo would be averted by the employer being motivated to take affirmative action steps in designing the algorithm rather than seeking to correct a disparate impact liability resulting from the application of the algorithm. By pursuing changes in the design phase of the algorithm, this affirmative action approach differs from an employer attempting to nullify or change an algorithm’s result as an ex post correction of any discriminatory disparate impact results. (Id.)
Ajunwa has also written about the impacts of technology in the workplace. But, her forthcoming paper addresses more specifically technology’s racial consequences. Ajunwa provides an interesting history of the ebbs and flows of labor actions throughout our country with any growth in improving working conditions for workers of color subsequently being met with newer ways to suppress opportunities and economic benefits for those workers. In cataloguing the historical developments from slavery to Jim Crow codes to increased incarceration to immigrant worker legal restrictions to a current lack of educational and professional opportunities, Ajunwa clarifies the depth of the plight faced by many workers of color seeking equality in our society.
After the illuminating discussion of the historical characterization of racist treatment of workers of color, Ajunwa transitions to the modern-day concerns of racism presented through technological innovation. Ajunwa condemns the expansion of digital platforms to employ gig workers classified as independent contractors who lack many of the legal protections provided to employees. Ajunwa also criticizes the use of algorithms in hiring especially through video interviews. According to Ajunwa, the door to discrimination opens “when these algorithms are trained on white male voices and faces, [because] they put applicants of color at a disadvantage.” Similar to some of the general scholarly concerns about racist algorithms, Ajunwa captures so clearly how design decisions about which inputs to consider can produce a neutral algorithm with racially-discriminatory results.
Ajunwa also discusses the concern that newer developments in surveillance technology create added burdens for workers of color who already have to find ways to “manage stereotypes attached to their race” in an effort to combat racism. With increasing surveillance creating key problems for the experiences of workers of color and the use of automated job tasks resulting in eradicating jobs typically employing those same workers, especially women of color, Ajunwa proposes a solution to rectify the racism produced by technological advances through any algorithms.
Similar to others, Ajunwa calls, in general, for broader and globalized labor protections to address the effects of advancing technologies on vulnerable workers. Though more specifically, Ajunwa ties the proposal to the needed protections for certain precarious workers including prisoners, undocumented immigrants, and guest workers who all tend to be people of color. Ajunwa also asserts that legislatures should conduct racial impact studies before implementing legislation that may create racial disparities. Also, Ajunwa identifies the need for greater economic stability for communities of color as a means of protection.
As a bottom line, Ajunwa beseeches us to not accept the digital growth as a futile development that will create a further racial divide. In this respect, both Bent and Ajunwa raise our consciousness to not only be aware of the racial disparities presented by employer use of technology and algorithms, they offer some methods to prevent this result. As a consequence, Bent and Ajunwa present us with some interesting reads while providing workers of color a potential respite from another racial hurdle in the workplace in 2021. These works also arrive at a time when racial unrest in our society has reached a boiling point and workers of color and their employers need some prescriptions to avoid race discrimination while moving forward in a positive manner.