Job Applications and Algorithmic Barriers to Your History

By James Myers

A friend (I’ll call her Mary) was laid off work last year because of a severe economic downtown in her industry. The difficulties she encountered in her job search, which lasted nearly a year before it met with success, are a warning for the preservation of accurate accounts of our life’s work.

The challenge is that many businesses now use algorithms to select job applicants for interviews. Since the algorithms look for characteristics that match the job posting, the successful applicant must be particularly good at using words and phrases that will satisfy the algorithm.

The result is not necessarily an accurate reflection of the human applicant, a difference the employer might later discover to be detrimental to job performance.

Mary gave me examples of algorithmic issues she faced in her long search. In one, she used the adjective “collaborative” in her resume, while the word “collaboration” appeared in the job posting. In the context Mary had used the word, it had the same meaning – but the algorithm didn’t understand the similarity of meaning where a human reader clearly would.

When Mary submitted her resume to one of many online resume analyzing services, the result wasn’t good: it was only a 40% match for the job description. With a number of wording changes, however, 40% turned into 100%. Mary’s problems didn’t stop there, however, because in several interviews she discovered that the intentions of the human who wrote the job posting were not properly understood by the algorithm that matched it to Mary’s application. As a result, while she fit the job as the algorithm interpreted the criteria, the human interviewer revealed very different requirements.

Image of manufacturing plant with vegetation inside the building generated by DALL-E

Mary’s experiences were not unique. Recently, the Washington Post reported on a 49-year-old job seeker named Troy George who “wanted to find a job as a manufacturing plant manager, so he went on several job sites including ZipRecruiter. But the site sent him notifications for secretarial jobs, janitorial positions and management roles at places with actual living plants.”

Job search sites now produce so many matches that both businesses and applicants often become overwhelmed with options, many of which turn out to be inappropriate.

The burden of sifting through such a huge number means good candidates can easily be overlooked.

With the goal of generating matches at greater speed than competitors, job search sites use a variety of algorithmic means which are not disclosed and can often result in biases that multiply over time.  As the Washington Post noted, “The sites’ algorithms factor in a candidate’s skills and experiences and, in some cases, jobs they’ve searched for, clicked on or applied for. Some sites’ algorithms identify best matches for employers based on previous searches, the type of candidates they usually seek more information from, and who previously did well. It’s unclear how much weight algorithms give to skills and qualifications compared with behavior on the site.”

In 2018, Reuters reported that Amazon abandoned an AI system used to recruit employees. The AI, which scored applicants in a range from 1 to 5 (like product ratings on Amazon’s website), turned out to be biased against female applicants. That’s because the AI had been “trained to vet applicants by observing patterns in resumes submitted to the company over a 10-year period. Most came from men, a reflection of male dominance across the tech industry.”

As I discovered in recently updating my own resume for the first time in more than a decade for a grant application, the software I used offered a menu of descriptions that I could have chosen to convey my work experiences.

I didn’t take the software up on its offering, for several reasons. For one, its descriptions were too generic, including the buzzwords that show up on so many CV’s they have become meaningless: “passionate and results-driven,” “motivated team player,” “detail oriented,” “dedicated professional,” and the like. Given a page to present my life’s story, I would rather it be a true representation of me. For another reason, in the limited space available I wanted to provide specifics of my accomplishments, not generalities.

Powerful AI tools like ChatGPT now deliver the ability to generate elegantly-phrased applications to match job requirements. Sites like My Perfect Resume offer “job-specific content, pre-written by experts” and promise that “We can get you hired 33% faster.” My Perfect Resume advises, “use Resume Customizer to see which keywords and skills your resume needs. It only takes a few clicks to customize your resume and improve your resume match score.”

The technological tools can, as Mary points out, help to empower the job applicant with repetitive tasks such as cover letters. ChatGPT also makes it easier to obtain information about the employer, and can help to create questions for the applicant to use in an interview.

The question becomes, for both employers and prospective employees, what is an accurate record of the history and development of a human’s skills? Is an AI qualified to assess a human’s qualifications and potential, for something so critical as an individual’s livelihood for potentially many years?