How HireVue is Encouraging Biases

If you’ve ever interviewed with Goldman Sachs, J.P. Morgan, or Vodafone, chances are you’ve experienced HireVue. 

HireVue is an AI-driven platform intended to help companies with pre-employment assessments via video interviewing and other challenges. It is currently used by more than 700 customers worldwide, including one-third of the Fortune 100. The algorithm used for video assessment analyzes facial expressions, voice inflection, and word choice, among other traits. It then compares these traits to the company’s employees who were previously hired for a similar role. One of the main goals of this automation is to “minimize bias in the selection process.”

Philip Brey, in “Values in Technology and Disclosive Computer Ethics,” talks about different types of biases-- preexistent, technical, and emergent. Preexistent bias, or an unconscious human bias against persons of certain backgrounds, is what HireVue is trying to eliminate. However, in doing this, HireVue has created a technical bias.

There are worries that the algorithm used by HireVue tends to favor white, male candidates, and that people with disabilities are likely to perform worse. Nonnative speakers, visibly anxious interviewees, and those who don’t fit the “ideal” model will also be unlikely to advance to the next round.

Brey claims that “embedded values are not necessarily a reflection of the values of designers.” The technical biases of HireVue may not be a direct reflection of the algorithm designers, but that does not mean that the humans behind this platform are free of all biases. In fact, I would guess that many of them are not. 

It seems as though, at least for now, algorithmic bias will continue to exist in HireVue and similar platforms. Instead of focusing on removing biases from these algorithms, it may be more productive to focus on removing these biases from ourselves. An AI-driven platform is a seemingly good solution to biased job hiring, but it also removes the need for individuals to address these biases face-to-face. Part of what makes us human is our ability to connect and learn from each other. By automating the hiring process, we are limiting our opportunities to do this.

James Moor claims “we are living in a period of technology that promises dramatic changes and in which it is not satisfactory to do ethics as usual.” At first glance, HireVue seems like an efficient way to remove biases, as well as save time and money throughout the hiring process. However, we also have to consider how this may be playing into the biases we already have, and preventing us from overcoming them.