🔒Is flawed job applicant software exacerbating the labor shortage?

The labor shortage in Connecticut and around the country is real.

According to a recent Connecticut Business & Industry Association survey of 625 local executives, 80% said they are having difficulty finding and retaining employees; 35% say the labor shortage is the greatest obstacle to growth.

The reasons for the shortage vary, but one issue that could be adding to the headaches is software used by many employers — big and small — to review job applications.

A Harvard Business School study released in September concluded the hiring process in the United States is “broken” and places most of the blame on software that screens out millions of otherwise qualified workers, often for trivial or arbitrary reasons, before a human being sees their applications. The result, the study concluded, is 27 million “hidden workers” who can’t find full-time employment because employers don’t know they exist.

“Ostensibly, automating hiring practices was supposed to reduce costs and ensure that companies found the talent to meet their current and future needs, while increasing diversity,” the study found. “But our research strongly suggests that the quest for efficiency in the hiring process has caused firms to narrow the pool of applicants they consider so severely as to exclude qualified workers.”

The study suggests that resume-screening computer programs may explain the post-COVID anomaly of companies struggling to hire qualified employees even as millions say they can’t find full-time work.

“Companies are increasingly desperate for workers,” the report says. “At the same time, an enormous and growing group of people are unemployed or underemployed, eager to get a job or increase their working hours. However, they remain effectively ‘hidden’ from most businesses that would benefit from hiring them by the very processes these companies use to find talent.”

A range of Connecticut academics and professionals say they agree with the study’s findings.

“It’s amazing to me how much money companies are spending on AI technology that is not only biased but broken,” said Yale School of Law Visiting Fellow Albert Cahn, executive director of the Manhattan-based Surveillance Technology Oversight Project and a critic of hiring algorithms. “It is actively getting in the way of hiring good job candidates.”

A technologist who has written software since he was 12 in addition to being a lawyer, Cahn said the programs are inherently flawed. First, the artificial intelligence and machine learning technologies that resume-sorting algorithms use will always produce bias. Second, these technologies rely on huge databases of past hiring decisions.

But why companies hire people and what they look for is constantly in flux, so the resulting software is always outdated, he said.

“When you look at how these systems are made, it doesn’t surprise me at all that they are making broad brush generalizations that are writing off millions of potential employees,” Cahn said.

Focus on job descriptions

Hiring software has transformed job hunting, according to Jill Koehler, Quinnipiac University assistant dean for career development, who said trying to find graduates jobs has gotten harder in recent years.

She said the hiring system is broken and algorithms are a big reason.

Much of her work with soon-to-graduate students now centers on teaching them how to get their resumes past algorithms to get interviews, she said. The key, she and other experts said, is identifying the words and phrases in the job description that the algorithm is programmed to flag and then incorporating them verbatim into a resume and cover letter.

To help her students do this, Koehler recommends a website called jobscan.com that provides tools and advice to help applicants crack the algorithms, which are called Applicant Tracking Systems (ATS). One tool, for example, has an applicant cut and paste his or her resume and a job description into side-by-side windows and then checks for matching words and phrases.

The site, which offers two free weeks and then charges $89.95 for three months or $49.95 a month, promises to teach users “how ATS impacts your job search and how to create an ATS-friendly resume that will land you a job interview.”

“It helps them to get a better understanding of how to optimize their resume,” Koehler said of the website.

Exacerbating the problem are poorly written, overly long and specific job descriptions that algorithms rely on, said Julia M. Fullick-Jagiela, a Quinnipiac University associate professor of management. The Harvard study highlighted the same problem, saying it creates a negative feedback loop where the more detailed the job description, the more qualified candidates the algorithm eliminates. Making matters worse, companies rarely update or rethink job descriptions beyond tacking on ever more finely-grained requirements, the study found.

“Often you see, when building job descriptions, they have laundry lists of qualifications that are not actually necessary for the job,” Fullick-Jagiela said. “I have horror stories of things that you’ve seen in job ads that shouldn’t be in job ads, that don’t make sense.”

Fullick-Jagiela said employers should prune back job descriptions and requirements to the absolute basics. She also strongly suggested having an employee who constantly monitors algorithms to assure they are not excluding qualified people or exhibiting bias.

“Is the algorithm working?” she said. “Do you need to tweak it in any way? It requires constant checking.”

What can employers do?

Anthony Avallone, owner of Reitman Personnel in Branford with 30 years experience in recruiting, said today’s AI-driven hiring system “dehumanizes the process.”

Many excellent job candidates, Avallone said, don’t grasp the role algorithms play in hiring and fail to change, for example, an unusual job title to something more generic. A simple edit like that can be the difference between getting through or not.

“I see candidates who are fantastic and they don’t get hired and they say, ‘Why not?’” he said. “Potential employers don’t know what they do.”

Avallone said the current reliance on software is a result of the migration of job ads to the internet. Employers that once received 10 or 20 resumes now got 100 or more. To work through the slush pile, companies turned to technology. That technology, however, has made it much harder for qualified people to get interviews, especially if they have gaps in their resume, a common reason why algorithms boot applicants, he said.

Local experts say hiring algorithms can be an effective part of the hiring process, but employers need to broaden their search beyond their typically narrow parameters. Koehler, who said one of her best hires didn’t make it through the screening phase, urged organizations to look for employees in other places like job fairs and chambers of commerce.

Avallone said companies should give applicants the opportunity to explain things like employment gaps instead of just screening them out and be more open to hiring older workers.

Fullick-Jagiela said organizations should focus more on job candidates’ potential instead of just past experience, as is inevitable with job-hiring software.

All three experts strongly agreed that companies should drop a hiring software-enabled fixation on finding the perfect employee with very specific skills and experience — a philosophy that eliminates many strong candidates.