Videos and A.I. in Hiring: Epic Fails or Helpful Tools?

Landing a job is not just a matter of mastering the technical skills required. It’s also about proving you’d be a good fit for the company culture. That kind of assessment is the reason for the job interview. In the past, these were all done in-person, with the company representative and the candidate meeting face-to-face in the same room.

With the advent of video technology, though, it’s become possible to overcome challenges of distance with virtual face-to-face meetings in live chat. Now, the next iteration of video interview is done without any live interaction at all. Instead, candidates record themselves on a video that will be reviewed by some person they likely will never meet… or perhaps not even by a person at all, but by an artificial intelligence (A.I.) agent. The last possibility raises serious concerns.

Possible EPIC Fail

The name in this space that has gained the most notoriety is HireVue, which conducts video recordings of job applicants for hiring companies. Those clients may include any number of things in the video assessment, from a test of soft skills to solving hard problems. However, there are concerns over how the platform allows A.I. to assess on-camera performances.

EPIC (Electronic Privacy Information Center) filed a Complaint and Request for Investigation, Injunction, and Other Relief against HireVue this past November. It charges that HireVue has failed to substantiate “that its technique meets the minimal standards for AI-based decision-making set out in the OECD AI Principles or the recommended standards set out in the Universal Guidelines for AI.” (The complaint does acknowledge that HireVue denies using facial recognition, though the people at EPIC don’t seem to buy it.)

Prepping for the Video Interview

In Asia, meanwhile, consultancies prep people to deliver what A.I.-powered video interviews are seeking. Reuters spoke with one of the consultants who does that in South Korea; he had hundreds of clients last year who came to him for preparation in what he called “the emergence of AI interviews.” These consultancies assume that the A.I. is reading their facial expressions, leading to guidance on how to frame a smile, for starters.

In the United States, meanwhile, there isn’t a prevalent culture of A.I. coaches. But according to a recent Inside Higher Ed article, we really should be paying more attention to the phenomenon: “Unilever, Atlanta Public Schools, Hilton Hotels and Resorts, and nearly 100 other employers now use HireVue, but little advice concerning HireVue interviews can be found on university websites.”

The article quotes Miguel Santiago, a senior at Baruch College: “‘No one is really sure what they look for when it comes to who moves on and who doesn’t.” All Baruch can offer students who want to prepare for such interviews is rehearsals via the Symplicity platform, watching their videos along with people from the school to discover what they need to work on. 

Much of the prep involved is not all that different from what one would do for any interview, whether there is a person present in the room or at the other end of a video link. The difference is that people feel a bit put off when they are talking to a camera with no feedback at all.

Expeditious (Not Necessarily Evil) Intent

The lack of feedback does make the “asynchronous video experience feel strange,” observes Steve O’Brien, President of Staffing at Jobs.com. He accepts it as an expedient option when the candidate pool grows too large and scattered to make in-person interviews feasible. It is not intended to filter out people on the basis of any legally questionable criteria, he maintains. Any attempt to program in such biases would, inevitably, come to light and come at great cost to the company.

He also believes that candidates would do better if they don’t approach such videos with the view that there is malicious intent behind them.

As for the arguments that videos can be used to reinforce bias in hiring, there’s the counter-argument that an A.I. is potentially just as biased as a human interviewer. If anything, an A.I. assessment can be more objective than a human, HireVue pointed out in Inside Higher Ed: “’Each algorithm or assessment model is trained not to ‘notice’ age, gender, ethnicity, and other personal characteristics that are irrelevant to job success, so it helps to level the playing field.’”

Using Video and A.I. in Different Ways

Used constructively, video can enable candidates to distinguish themselves by sharing their story; the underlying A.I. technology can filter these candidates to a job that ideally suits them.

In those instances, A.I. can also focus on résumés, making what O’Brien calls “contextually smart” job recommendations. That means that a skill that is not listed explicitly can be inferred from contextual details within the document. Further contextual analysis can derive the difference, for example, between “a tech cloud engineer that built up the infrastructure for a bank versus one who built an ecommerce solution.”

“We look at three things: skills, capability, and fit,” O’Brien explained. “Skills are what people know how to do. Capability is what they are born to do.”

To illustrate, he said there could be an engineer in possession of the skills listed for a corporate job who won’t be happy there if his own work style makes him better suited to a startup. Candidates can take an assessment in which they “identify what gives them greater satisfaction and lower stress.” Job applicants also reveal more about themselves over time in the way they interact with a platform, which can show if they really are more amenable to structure and consistency or prefer constant change.

Technology is a tool, and both A.I. and video can prove to be helpful or harmful, depending on the aims to which it is applied.