The new recruitment challenge: Filtering AI-crafted résumés

Candidates are using generative AI tools to build résumés that are customized to each job for which they are applying. They then can fire off résumés in rapid succession or have bots do so for them. They are also relying on AI to prepare for interviews, which is leading to a sameness among applicants.

As a result, tech firms and recruiters are reinserting some human review and adding screening steps.

While candidates have long inflated their experience during their job searches, recruiters say there has been a rise in so-called fake candidates—people who misrepresent their experience and identities. These candidates are applying for remote roles with experience on their résumés that is later found to be made up. Recruiters suspect fake candidates are getting through Zoom interviews either by using software or having a third party feed them answers.

For years companies have been relying on software that automates various aspects of hiring. The process has led to frustration among applicants who can be removed from consideration without their résumés ever being viewed by a human. The combination of a cooler labor market and the ubiquity of remote interviews is only exacerbating the onslaught of AI-related issues.

“Nearly all of our customers, particularly the larger tech companies, are struggling mightily with the amount of applicants they’re getting," says Ben Sesser, chief executive of the interview-software platform BrightHire. “Our clients are noticing the use of AI a lot throughout the hiring process."

At Lattice, a human-resources software platform, the talent team is leaving job postings up for only 24 to 48 hours because of the volume of résumés. Mendy Slaton, director of people operations at Lattice, attributes the influx of applicants to the tighter labor market and the ways AI is expediting the application process for candidates. What used to take hours or weeks can now be done in minutes, says Slaton.

Lattice is considering additional steps in the application process. There are optional questions, for example, asking candidates to list three reasons they are interested in the role, before submitting their résumés. The company is thinking about making that step mandatory as an additional filter.

“I want to use AI to help me speed up that process, but I don’t want it to replace the human judgment pieces," says Slaton.

Extra vetting

In some cases, the solution seems to be more tech: BrightHire’s software records interviews. Several clients have told Sesser that in reviewing footage they have noticed interviewees looking away from the camera before answering a question, a sign they might be using ChatGPT and a prompt for further vetting.

Some companies are requiring job seekers to record a video of themselves or answer preliminary questions before their submission is considered complete to mitigate the volume of inbound applications and weed out those who might not actually be that interested in the role.

Many tech jobs require applicants to take tests that require them to showcase their coding skills. While the use of ChatGPT for coding tests isn’t prohibited by many companies, some executives have started requiring candidates not only to take the test but also to show their work during a live interview and explain what tools they used and why. News Corp, owner of Dow Jones Newswires and The Wall Street Journal, has a content-licensing partnership with OpenAI, the creator of ChatGPT.

The current hiring landscape is a stark departure from the many years tech companies emphasized growth and couldn’t bring on talent fast enough. These days, tech companies are focused on efficiency, leaner operations and putting money toward AI investments.

Tech talent remains core to the strategic plans of most companies, and unemployment in the industry is lower than the national rate, according to the tech trade nonprofit CompTIA. But many workers are still job hunting after being included in layoffs.

Matt Somero, a recruiter who works on a contract basis in-house with two seed-stage startups, says that when he posts jobs for those companies, he receives several hundred résumés for each role within the first 48 hours and then has to close out the posting. On average he follows up with only 10 to 15 people.

Somero keeps coming across candidates who seem to be throwing spaghetti at the wall, applying to a senior role with only two years of experience, for example.

Automation tools “appear to be enabling mass and fairly indiscriminate applying," he says. “There are really good folks that are just not getting seen because of the volume."

Somero also has encountered applicants who he determines through online sleuthing aren’t who they claim to be. He has been on video calls in which candidates have blurry backgrounds and sound as though they are in a call center. Somero has started checking university alumni directories to verify whether applicants went to the school listed on résumés and checking out profile pictures for what he considers to be red flags, such as a photo taken from a far distance.

Chris Abbass, CEO of the recruiting firm Talentful, says he knows of companies that have accidentally hired “fake candidates" for remote roles. It became clear, he says, that the employees had lied about their experience when they weren’t able to do various aspects of their jobs or when someone showed up on a video call who didn’t match the appearance of the person who was interviewed.

“There is software that people can use on a phone screen where they’re playing the software in the background, and it’s feeding the candidate the answer," he says. “There are household names that have hired people and then realized a couple weeks in that this person isn’t the same person."

ChatGPT effect

Arun Saigal, CEO of the mobile app development platform Thunkable, says he is finding lately that some job hunters are listing generative AI experience on a timeline that doesn’t make sense unless they had worked for just a few companies.

“You may be smart, but you just don’t have that experience because that’s not really possible," he says.

Saigal requires a coding test for some roles and doesn’t mind if candidates use ChatGPT. They are allowed to use any tech that is helpful, but they are expected to disclose it. “We ask you directly, ‘What tools did you use?’ " he says.

Candidates are then required to walk through their test and explain the thinking behind the coding choices they made during a live interview.

“That’s how we ensure you didn’t just take something from ChatGPT and just copy and paste and have no idea what you did," he says.

Satyam Dave, a 20-year-old student at Purdue University who is majoring in computer and information technology, has leveraged AI in his job search and has encountered a potential employer’s efforts to outsmart it.

Dave created his own AI-based résumé-building app that uses keywords from job postings to optimize a candidate’s experience. He had to record himself in a one-sided video interview in which he had up to 30 seconds to prepare and answer each question.

Daniel Lee, a data scientist who used to work at Alphabet’s Google and now coaches tech workers, finds his clients are using AI more during his mock interviews with them. He says he can tell from the uniformity of their answers.

“It’s a little bit awkward," Lee says. “I don’t want to call them out and say, ‘Are you by any chance using ChatGPT?’ "

Write to Katherine Bindley at katie.bindley@wsj.com