Kendall McGill had only been on the interview call for a few moments when she realized something wasn’t right.
“It was a standard AI voice,” she says. “When you get on an interview, you can tell if it’s a real person.”
She had applied for a project management position, but wasn’t told in advance that the interview would be conducted by an AI agent. She started to answer the AI agent’s question, but then decided to hang up.
“I can’t say that I felt upset; It just made me uncomfortable,” says McGill, who lives in Baltimore. “I would much rather talk to human beings and get those experiences that you get when talking to a human.”
McGill isn’t alone. A growing number of job seekers are finding themselves being interviewed by artificial intelligence, sometimes without warning. Companies are increasingly using AI to help sort, schedule, and screen applicants in a race to reduce hiring costs and time. But for many candidates, the experience feels alienating and impersonal.
Companies are turning to AI to pick the best candidates
According to a recent report by Resume Now, a resume building platform, 96% of the more than 900 U.S. hiring professionals surveyed said they used AI in recruiting tasks, such as screening and resume analysis. While the survey didn’t ask respondents if they used AI specifically to conduct interviews, about 94% of those surveyed said AI screening tools effectively identify strong candidates. Roughly 73% said using AI tools has sped up the time-to-hire.
Career experts say it’s not a surprise that companies are embracing the new technology—it’s typically faster and more cost-effective than hiring recruiters. But some are warning that over-relying on AI could introduce bias and undermine the candidate experience.
“It’s important to make sure that AI or automation doesn’t completely take the human element out of the hiring process,” says Keith Spencer, career expert at Resume Now. “You need to consider the impact to job seekers when they are constantly dealing with AI instead of a real person.” He adds that the candidate experience is often the first glimpse into a company’s culture and could ultimately affect retention.
A surprise on the other end of the call
Wafa Shafiq, who lives in Mississauga, Canada, was also caught off guard when she interviewed for a marketing specialist role in the fall.
After submitting her application through a recruiting agency, she soon received a request to schedule a video interview. The confirmation email said she would be speaking with “Alex from Apriora.” A quick search revealed that Alex wasn’t a person—it was an AI recruiter.
In the end, the interview “wasn’t as bad as I thought it would be,” Shafiq says. The AI agent asked her about six or seven questions, and gave her time to ask her own. It could handle general queries, but struggled when she asked anything specific or nuanced. The experience, she says, was “cold, but efficient.”
“There was no small talk. There was no chance to connect in the way that we do with a recruiter, which was a little bit odd,” Shafiq recounts. “It felt a little dystopian, but felt very efficient because it was very to the point.”
Shafiq didn’t hear back about the job. While she’s open to interviewing with AI again, Shafiq says the process would have felt far better with some form of human interaction. Even if, she says, it were just to provide a “heads up, this is going to be an AI interview, here are ways to prep—like that just makes me feel the company is setting me up for success more than like, ‘sign up for a slot in this AI interviewer schedule and just show up.’”
Spencer from Resume Now agrees that interviewing with an AI agent “should not be a surprise.”
“You shouldn’t enter into an interview fully thinking you’re going to be speaking to a real life human being, and then all of a sudden, you’re interacting with an AI avatar,” he says. “That just goes back to this idea of transparency—if you’re going to use AI in that way, then the candidate should be fully aware.”
What happens after the AI hangs up?
Matthew Bidwell, professor of management at the Wharton School of the University of Pennsylvania, says that using AI to conduct interviews resembles older, one-way video interview tools. Those recordings have usually been reviewed by a human. But now, Bidwell worries that hiring managers could use AI to both screen and evaluate candidates, which he says “feels quite alarming.”
“The early work on these large language models does suggest that they do have some kind of race and gender bias baked in,” Bidwell says. “It creates a bunch of legal and ethical issues.”
Some studies, including one conducted by Algorithmic Justice League founder Joy Buolamwini, have uncovered gender and racial bias in some AI systems.
Candidates are using AI too
Eric Lu, co-founder of the video editing website startup Kapwing, had a surreal experience interviewing a candidate who was applying for an engineering role at his company. At first, the interview was going really well, Lu says—the candidate seemed like a good fit for the role and liked Kapwing’s work. But as the interview went on, the candidate struggled to answer certain questions.
“What they were saying about the technology started to not add up with what was happening in what would be a reasonable, real life situation,” Lu says.
The candidate said they worked on a daycare-related app, and had developed a feature that helped send out text messages to all the parents. When they said they weren’t able to send 30 texts at a time without running into challenges in the coding, Lu asked them why they would need that feature and what those texts were about. The candidate couldn’t answer the question.
At another point, the two discussed “lazy loading,” which is when part of the content on a page loads faster for the user and the rest of the content loads later as the user is interacting more with the page. Lu asked the candidate what content they loaded faster on the app. Once again, the candidate was unable to answer.
“I think that’s a pretty straightforward question to answer if you actually worked on it,” Lu says. “It was crazy that basically this candidate just went completely silent.”
Lu confronted the candidate. Eventually, they admitted to using AI to prepare for the interview.
“This was a candidate that had a real LinkedIn profile, seemed to be a real student at a reputable college, had real job experiences under their belt,” Lu says. “But [they] still chose to prepare for this interview in this way.”
Lu wasn’t sure if the candidate was using AI live during the interview, but suspects the candidate used AI beforehand to generate answers that they then memorized, and then when Lu’s questions were out of the bounds of those responses, the interview broke down.
Bidwell says that candidates reading from an AI-generated script for an interview “is hugely problematic” for companies. A solution, he proposes, could be to hold interviews in person.
Lu’s team at Kapwing had come across AI applications before that interview, and had set up multiple internal guardrails to weed out “bot” candidates or ones that rely too heavily on AI—they go through applicants’ online presence to make sure they’re a real person, check their work experience to make sure they worked for real employers, and require live video calls for interviews. But he says that interview experience has raised even more questions for him about the hiring process.
“What does our interview process look like in the age of AI? And how does that have to change?” he says. “I think there’s a lot of questions there.”
Resume Now’s report found that 79% of hiring professionals who were surveyed said that there should be some sort of regulation over the use of AI-generated content in job applications, indicating that while employers are increasingly using AI in the hiring process, they are also concerned about applicants using the technology. While that can feel “a little contradictory,” Spencer says, he can understand “both sides of that coin.” What is key, he continues, is that both employers and candidates use AI ethically—in other words, to support or enhance their work, not replace it.
Read More: In the Loop: Is AI Making the Next Pandemic More Likely?
Lu says Kapwing hasn’t invested in using AI to conduct interviews, but that he understands why companies would, calling the use of AI in the hiring process “a tricky situation.”
“If I was a candidate, I would be pretty sad about it,” Lu says. “But as somebody who is running a business that may get dozens or hundreds of applicants a day, I understand the need from a business perspective of needing to screen the candidates; you just don’t have enough manpower hours on your own team to put in that human touch.”
But for people like McGill, it’s that missing human touch that makes all the difference.
“I think there are certain auditory or visual cues that you get when you’re talking to a real person that is just hard to communicate strictly with a computer,” she says. “And I just did not want to do it.”
The post When Your Job Interviewer Isn’t Human appeared first on TIME.
