Inhuman resources
Efficiency-obsessed hiring managers are farming out inter views to robots. But job seekers miss the human touch of a real conversation
Afew weeks after Eden Hoffman applied for a project management role at a bank, her inbox pinged with a message inviting her to an interview. Hoffman’s excitement quickly gave way to confusion, however: Rather than meet with a potential manager, the 23-yearold was asked to field questions from a robot, which would record her answers through a webcam and use an algorithm to score her performance.
Hoffman had interviewed for jobs remotely before, but those interactions were over Skype — with sentient humans. “I had never heard of a one-way video interview that records you,” she tells The Post.
Firm handshakes and a snazzy interview briefcase matter less and less in the digital age, when the all-important first impression you make could be on a machine. Hiring managers have long used digital tracking systems to scan for keywords on résumés; now, they’re enlisting even smarter bots to conduct interviews on their behalf. In a December 2017 survey by San Francisco tech firm Entelo, 62 percent of busi- ness leaders said they plan to invest in artificially intelligent tech engineered to analyze human behavior and pinpoint top talent. Springing up to meet that need are robo recruiters such as HireVue, Mya, Sonru and Spark Hire, which have already led staffing efforts at major employers, including Geico, Oracle, MercedesBenz and Tiffany & Co. Proponents say AI-driven hiring cuts bias and improves efficiency. Skeptics say it rubs them the wrong way.
For her interview, Hoffman staked out a plain white wall in her Gramercy apartment and fussed over staging. “I had to make sure the sunlight from the window wasn’t obstructing the picture, and I took off my glasses in case of glare,” she says.
She put on a black blouse and blazer. After checking to make sure her lower half was out of frame, she decided to leave on her workout leggings.
The questions were fairly standard, focusing on her work experience and skills. But Hoffman says she felt uneasy during the call.
“It was very unnatural,” she says. “I’m a big people person; when I’m in front of someone, I’m using my hands and making eye contact. It’s awkward when you’re just doing that for yourself.”
Those physical cues count, even when you’re talking to a video bot. Loren Larsen, HireVue’s chief technology officer, tells The Post that his company’s AI platform can analyze up to 25,000 speech and behavior variables, including vocal pitch and smile frequency, in calculating a candidate’s overall score.
The algorithm is tailored to the employer’s needs.
“For example, a person who smiles frequently may be scored highly in an interview for a customer-facing role,” Larsen says.
What you say matters, too, Larsen says, as the HireVue bot can be “taught” to favor candidates who use desirable word clusters and demonstrate certain verbal patterns.
“[Job seekers] may talk in very abstract terms, or they may speak in concrete terms,” Larsen says. “If I need someone who is a leader or a philosopher, maybe I want someone who’s more abstract. If I need someone to build widgets, I don’t want someone who’s like, ‘Theoretically, possibly, that could happen.’ No, I need things that will actually get done. [Those candidates] tend to use more active language.”
Still, even robots can perpetuate bias when an algorithm isn’t nuanced enough. Amazon reportedly had to retire an experimental AI recruiting system in 2015 when the bot gave disproportionately low scores to female candidates. Apparently, the platform had been programmed to favor job seekers whose credentials resembled those of the company’s star performers, most of whom were men with backgrounds at male-dominated companies.
Despite the vast inputs that today’s robots can analyze, New York City-based career coach and management consultant Avery Roth says AI recruiting leaves humans at a distinct disadvantage.
“It narrows the scope of how a candidate can connect with a company and therefore position themselves in the best way possible,” she says.
Robo chats also hamper the job seeker’s ability to ask questions and come to their own conclusions about whether a position is a fit, she says: “A lot of important information comes through during [a human] exchange.”
But Larsen says the pros outweigh the cons, and that machine learning makes it possible to forecast with startling accuracy who will thrive in a given role.
“We can tell our customers who is most likely to be successful,” he says. “We look at what [the candidate] said, how they said it, what the nonverbal cues were, what emotions were being conveyed. We’re validating competencies — like negotiation skills and confidence — scientifically, against performance measures.”
Rather than dehumanizing, Larsen says robo interviews are authentic, letting concrete credentials shine through.
“The algorithm pays attention to what actually matters,” he says. “A hiring manager will say, ‘I just felt comfortable with that person’ or ‘She’s attractive.’ That is not the right way to hire people.”
Roth still prefers a traditional approach and advises her clients to circumvent AI interfaces to whatever extent possible.
“Network your way to the people who are in a trigger-pulling capacity,” she says.
And if there’s no getting around a robo-gatekeeper? Act as natural as you can during your on-camera cross-examination, then do your best to connect with a living being.
“Reach out to a person in the company so they know your video is coming through,” Roth says.
Although Hoffman impressed her robot interviewer enough to score an in-person meeting at the bank, she didn’t walk away with an offer. Instead, she landed her current gig as a policy communications analyst the oldfashioned way.
“I met someone face-to-face, we sat down, and it led to a job,” she says.