We Asked ChatGPT to Spot Red Flags on Resumes. Here’s What Happened.

Spoiler alert: Recruiting still needs humans.

Article main image
Aug 10, 2023

Research shows that 79% of recruiters believe that “people won’t be involved in recruitment” in the future. Additionally, 56% of people think hiring decisions should always be done by a human.

If U.S. Senators Bob Casey (D-PA) and Brian Schatz (D-HI)’s new “No Robot Bosses” act becomes law, humans will have to stay involved. “As robot bosses become more prevalent in the workplace,” Casey said, “we have an obligation to protect working families from the dangers of employers misusing and abusing these novel technologies.”

With so many recruiters believing they can be replaced, I decided to look into what today’s AI technology looks like when it evaluates resumes. This is especially keeping in mind that just five years ago, Amazon stopped using their custom AI technology because it discriminated against women.

Has AI gotten better? I asked professionals to run their own resumes through ChatGPT and ask it to identify potential red flags. The results indicate that perhaps it’s not time to turn over hiring to the bots quite yet. (To protect these individuals so that when people Google “resume red flags” their names don’t come up, I’ve kept them anonymous.)

Judging Time Frames

One HR expert has been in her current job for close to four years and was at her previous job for almost seven years, and the job before that for eight years. That seems to indicate stability, but ChatGPT raised her short tenure at jobs in 2007 – 2008 as a red flag. It identified this red flag:

Frequent job changes in early career: The candidate seems to have changed jobs relatively frequently early in their career, with several positions held for less than a year. While this may not necessarily be a red flag, it could raise questions about job stability and commitment.

Another professional had four jobs in 18 years and got a similar flag: The candidate has had several job changes throughout their career, which might raise questions about stability and loyalty.

Understanding Multiple Roles

While it’s not uncommon for people to hold multiple positions, especially in today’s world of side gigs, ChatGPT flagged an academic for what looks like success.

Overlapping roles: There are instances where the candidate held multiple positions simultaneously, such as being an Associate Dean while also serving as a Provost Fellow. This might raise concerns about time management and potential conflicts of interest.

The fact that this person served in multiple roles successfully should indicate good time management skills instead of being a red flag

Recognizing Transferable Skills

An HR professional shared this red flag:

Unrelated past role: The candidate’s work experience includes a significant period spent as a case manager for mental and behavioral health services, which may not directly align with their current HR career path. Employers might inquire about the transition and continuity in their career trajectory.

The HR professional summed up the problem with this commentary, “This is why AI will never replace humans. If you can’t understand how a decade in the mental health field aligns with HR, I’ve got some news.”

Even if wanting to change career paths were a red flag, this HR professional is currently in an HR role. It makes no sense to say staying on the current path is a red flag.

Old Employment Gaps

While many hiring managers and recruiters consider employment gaps to be red flags, one from 2014 should not be a concern. Yet, Chatgpt flagged it:

Employment gap: There is an employment gap between October 2014 and January 2015, which is not explained. While a short gap might not raise major concerns, an unexplained extended gap could be a red flag for employers.

Not only is it an old gap; it was for a maximum of four months. If a candidate is laid off, that’s a very respectable amount of time to find a new job.

Flat-Out Mistakes

ChatGPT hallucinates and misunderstands things all the time. Therefore, it’s not surprising that it misread a resume. A project manager shared this red flag:

Employment gap: There is a gap between the last role in November 2013 and the subsequent role starting in April 2013. It’s essential to clarify the reason for this gap during the interview process. Frequent job changes: The applicant has changed jobs multiple times in the past few years, which could raise questions about their stability and commitment to long-term positions.

As the project manager clarified: “There was no gap. ChatGPT read the dates wrong, and the job changes were promotions at the same company.”

Many recruiters and hiring managers want to see promotions at the same company. That shows not only an increase in responsibility but stability and commitment to an organization, both of which are valued.

Simply put: ChatGPT misread and then misevaluated what that meant.

Does this mean ChatGPT is not helpful in evaluating resumes? ChatGPT can certainly summarize and point out potential positives and negatives, but as in all ChatGPT functions, it takes a human to evaluate the results. In other words, recruiters aren’t going anywhere any time soon.

Get articles like this
in your inbox
Subscribe to our mailing list and get interesting articles about talent acquisition emailed weekly!