A firm handshake used to be all a job candidate needed to make a good first impression, but today a growing number of firms are relying on artificial intelligence to screen applicants.
Using AI to conduct first-round interviews saves time and money. But it also changes how job candidates act — and not always for the better, according to a new study from the University of Georgia.
The researchers found job seekers often overcompensate for nervousness during one-way interviews when they know they'll be evaluated by AI. And AI candidate rating systems can't tell when candidates are stretching the truth.
"We spoke to a lot of people who were interviewing with AI recruiters, and it was affecting the way they would behave during the interview," said Akshat Lakhiwal , corresponding author of the study and an assistant professor in UGA's Terry College of Business . "So naturally, that affects the outcome of the interview."
Job seekers more likely to exaggerate, embellish when interviewing with AI
During a "one-way" asynchronous video interview, a job seeker records answers to basic interview questions such as "Why should we hire you?" or "Can you tell us about yourself?" When the practice was first introduced, the tool made it easier to schedule initial interviews, and the sorting process made it easy for recruiters to compare how multiple candidates answered the same question.
Today, many companies save time by allowing AI systems to grade and sort candidates' responses. The change has left candidates feeling somewhat disoriented and helpless, Lakhiwal said.
"If you're applying for a job at your dream company and your dream company wants to interview you using AI, you really don't have a lot of choice," he said. "You don't really want to say no to it, but you also don't know how it works."
That leads many applicants to change their behavior to fit whatever they imagine the criteria might be.
A lot of embellishment was happening. (Applicants) seemed to be throwing the kitchen sink at the situation to try to give the "evaluator" what it was looking for.
Akshat Lakhiwal, Terry College of Business
In a study of hundreds of online job seekers, Lakhiwal found a considerable increase in applicants' self-reported "exaggerations" or "deceptive embellishments" when candidates knew their videos would be evaluated by an AI agent rather than a human. An analysis of their videos and answers confirmed what they reported.
When participants were asked whether embellishing qualifications was ethical, they said it was necessary to perform well in an unpredictable environment.
"A lot of embellishment was happening," Lakhiwal said. "They seemed to be throwing the kitchen sink at the situation to try to give the 'evaluator' what it was looking for."
On the hiring end, the industry-favored AI agent Lakhiwal used in the study didn't penalize job applicants who stretched the truth and scored them as well as those who earnestly described having the same qualifications.
When human evaluators saw the videos, they seemed able to discern this behavior. They generally penalized those candidates and gave their highest ratings to those who seemed to engage in more authentic behaviors.
Explaining how AI will evaluate candidates improves interview performance
Lakhiwal's team found a way to curtail candidates' tendency to exaggerate for AI evaluators was to explain the interview process.
In an experiment, one group of interviewees was told their videos would be reviewed by AI and then given specific details about what the AI was looking for.
They were told the system would look for facial expressions, verbal sentiment and specific keywords, and that it would rate their videos based on teamwork, job-related abilities, work style and personality.
This group reported and displayed the same amount of authentic behavior as those being told they were being reviewed by a human.
Telling applicants more about the process allows them to be more authentic.
Akshat Lakhiwal
"Traditionally, companies have refrained from transparency in the hiring process," Lakhiwal said. "They don't like that word because they feel if participants know how they will be evaluated, the applicants may game the system. And there is research that has showed this. But here we found that telling applicants more about the process allows them to be more authentic.
"We're advocating transparency in terms of someone's ability to understand the process. I don't really need to know which model they are using to analyze my video, but as a candidate, what matters is to be able to make sense of the process the same way I understand what it means to be interviewed by a human."
Understanding how humans interact with and react to automated systems in the workplace is key to keeping these types of technologies from backfiring on organizations.
Published in Information Systems Research, the study was co-authored by Che-Wei Liu of Arizona State University, Hillol Bala of Indiana University and Hung-Yue Suen of National Taiwan Normal University.
