When your job interviewer’s initials are AI

Everybody worries that smart computers will take over jobs and get people fired. What’s less visible, but perhaps equally important, is that they’re also playing a role in who gets hired.

Artificial intelligence is spreading into the hiring process despite widespread public skepticism. Innovations in automated interviewing are happening so quickly that governments are struggling to keep up with ground rules. Left unresolved are larger ethical questions, such as: Can we trust machines to judge human talent and potential? Will AI interviewers be less biased than human ones?

Such questions haven't stopped companies from rushing to embrace the technology, at least for entry-level positions and internships, despite drawbacks of machine-based interviewing and previous problems with discrimination. 

“I think 100% of our students will encounter AI whether they know it or not” in the interview process, says Steve Rakas, executive director of the Masters Career Center at Carnegie Mellon University’s business school in Pittsburgh.

Among employers considering the technology, “it went from fear to FOMO, meaning fear of missing out,” says Patrick Morrissey, chief customer officer of HireVue. The talent-search technology firm in South Jordan, Utah, counts 60% of Fortune 100 companies and 13 federal agencies as clients.

For large organizations, the advantages are obvious. Swamped with applicants for beginner-level positions, large companies are eager to streamline the hiring process. For example, Goldman Sachs says it is now using AI to help screen the tidal wave of résumés it receives for its high-paying, entry-level spots. (In 2022, the bank says, it received 236,000 applications for 3,700 positions in its internship program.) 

Small companies can benefit from artificial intelligence assistance, too.

In 2016, Michelle Zhou and her co-founder at Juji, an AI hiring startup in San Jose, California, spent 2 ½ weeks sifting through 700 applications for one intern position. Two years later, having molded their AI analytics into a full-blown, automated interviewing system, the pair used it to judge talent and pick candidates for follow-up interviews. The process was finished in a couple hours. 

Juji’s text-based technology not only speeds up evaluation of candidates, but also is less biased, Ms. Zhou contends. Before the AI interviewing system was in place, she and her colleague used shortcuts to sort candidates, such as the reputation of the applicants’ alma maters. When they switched to the AI system, it chose “one of the people we would never have picked up, because they came from one school we had never heard of,” she says.

Several universities, already training students for live job interviews, now help them handle AI video interviews, too. 

“The tricky part of the AI interviewing [is] a timer,” says Jeanine Dames, director of Yale’s career strategy office. So as a candidate is answering something like “tell me about yourself,” they are also watching the time tick away. When the counter reaches zero, the recording stops. 

“It is really awkward to sit there and talk to yourself and a computer,” says Hank Michalik, a Yale student who had an AI interview for a finance internship last spring. “It is dehumanizing.”

The computer systems evaluate various skills of candidates, such as their attention to detail or their ability to work with others. If a company is looking for multitaskers, HireVue can introduce a game during the interview in which, while answering questions, the candidate has to watch a volume indicator on the monitor and click when it falls below 90%. The system is measuring what the company calls “human potential intelligence.”

“It’s not about the degree that I got 30 years ago in university; it’s about what am I capable of,” says HireVue’s Mr. Morrissey. “So eliminating degree requirements and just focusing on skills and capabilities and competencies is very much where the world is going.” 

While AI could well prove to be less biased than human interviewers, it’s not foolproof. 

Amazon disbanded an AI résumé-screening system in 2018 after it found the technology discriminated against women. Last September, iTutorGroup agreed to pay $365,000 to settle a federal lawsuit claiming its system automatically rejected older applicants for tutor positions. 

In 2021, the federal Equal Employment Opportunity Commission began monitoring AI interviewing systems for bias and last May issued a technical document to help companies avoid job discrimination. And this past July, New York City passed a law requiring companies to perform annual audits of their AI hiring and promotion systems for bias. In October, President Joe Biden issued a wide-ranging executive order on AI that called for principles and best practices to keep companies from “evaluating job applications unfairly.” 

While the automated hiring systems have relied on other forms of artificial intelligence to analyze candidates for some time, the arrival of generative AI a year ago has accelerated innovation in interviewing technology. In December, for example, Juji unveiled a system that automated the interviewing of references given by candidates. The system not only judges whether the references offer an unbiased view of the person, but also summarizes all the interviews into a single document for companies to assess.

Now, a couple of Juji’s clients are trying out systems that would use the technology not just for entry-level positions, but also for higher ones. Companies emphasize that humans, not machines, make final hiring decisions.

The public remains skeptical of the technology. By a 71% to 7% margin, respondents to a Pew Research Center survey last spring opposed using AI to make final hiring decisions, and 41% opposed using it at all to review job applications. A September poll for the American Staffing Association found that just under half of employed people looking for a new job thought the technology would be more biased than human interviewers.

“I think the positives outweigh the negatives in the big picture,” says Ms. Dames of Yale. “It’s easier for a large organization to recruit from 200 universities around the world, even if they can’t visit 200 universities. [And] a small organization could use a large screening tool and invite applications from thousands of students. ... It gives so many more people the opportunity” to be considered by quality employers.