November 11, 2019

LEE | The Disturbing Reality of One-Way Video Interviews

Print More

As the fall semester begins to wind down, fall recruitment season also nears its end. For some, this could be a time of joy as they receive offers to their dream company, while others continue the search in hopes of having better luck with the next cover letter they submit. Because of such a focus on the outcome of the recruitment cycle, both candidates and employers appear to have less consideration for the process through which an offer is extended. Firms rarely ask for feedback regarding their process despite many candidates having strong opinions about a particular employer’s recruitment method.

One particular practice in the candidate vetting process has been particularly off-putting: one-way video interviews. This type of interviewing now has widespread use among large firms seeking to minimize scheduling complications with interviewees, increase efficiency with candidates and remove potential interviewer bias. Prominent employers such as Goldman Sachs and Unilever use artificial intelligence to screen candidates. Many college students like myself at Cornell and elsewhere are prompted to speak to their computer screen with no human presence at the other end of the interview. When an interviewee submits their video interview to platforms like HireVue, they are assessed not only based on the content of their responses but also on emotion detection cues that compare them to “top performers” at the organization. HireVue contends that their assessment tool eliminates human bias and can actually increase employee diversity.

Yet, such AI-based evaluation platforms continue to leave questions unanswered due to the proprietary nature of their algorithms. Firms like HireVue are permitted to veil the exact standards used to evaluate candidates, leaving much room for ambiguity. These tools create models based on established practices that, despite intentions to remove human bias, could simply reproduce the demographics of existing staff. This is particularly alarming when only 18% of the tech workforce is female in major technology firms like Google. And because AI technology is only as thorough as the input it uses, certain interviewees could be viewed as favorable for possessing traits traditionally considered masculine. The algorithms written by humans from male-dominated fields like tech could simply create a ‘closed-loop system’ that systematically discriminates minority groups, as claimed in a class action lawsuit against Facebook. What’s more, the algorithm doesn’t provide assessment scores or explanations to its decisions, unlike a human recruiter.

But there are indeed benefits to these assessment tools. Designed to minimize human subjectivity in the interview process such that candidates are not evaluated based on arbitrary traits, the metrics used are often considered more objective than those by people. For instance, Unilever found a 16% increase in diversity hires through the HireVue platform they utilized, which enabled them to diversify the talent pool.

However, these one-way video interviews still ignore many other aspects integral to the recruiting process. Candidate experience is easily neglected as companies use third-party programs that are impersonal and disconnected from company values. Responses in one-way interviews are cut off after a set time limit, disempowering candidates even more than they already are. Interviewees are not given the chance to ask questions or demonstrate themselves in ways outside of the predetermined questions, posing a challenge to exemplifying how they are not simply one face among many others. And despite the rampant potential for misuse of a well-intended tool, companies continue to use inherently biased algorithms to discern one applicant over another.

Firms can still incorporate the merits of technology in other ways that are beneficial to both parties. For instance, skill-based assessments conducted online enable the firm to evaluate the applicant in a timely fashion and on an objective set of criteria such as their ability to process Microsoft Excel functions. This allows the organization to continue screening candidates and ensure that prejudiced factors like those corresponding to facial expressions don’t select one equally qualified applicant over another. One-way video interviews shouldn’t be the main standard for rejecting or moving candidates forward. Candidates’ entire profile should be taken into consideration through a multifaceted approach that blends resumes, neutral screening software and interviews.

As companies seek to attract the best and brightest applicants from a diverse talent pool, they must think once more about the implications of the technological tools they utilize. In an era where Facebook likes can be utilized to predict gender, sexual orientation or even political beliefs, using AI-powered technology with caution to determine something so significant as interview results has become important now more than ever. As someone preparing to enter the workforce, I hope that my skills are not determined by an artificially created model that rationalizes the bias it was built upon.

DongYeon (Margaret) Lee is a senior in the School of Industrial and Labor Relations. She can be reached at margaretlee@cornellsun.com. Here, There, Everywhere appears alternate Tuesdays this semester.