The automated race to use artificial intelligence to screen job applicants

Artificial intelligence increasingly is becoming a go-to tool for human resources departments in the hiring process, even as some professionals raise concerns about potential bias and reliability of the software programs.

With the growth of online job boards and networking sites such as Indeed, Monster.com and LinkedIn making it easier for job hunters to submit resumes, overwhelmed HR departments are using artificial intelligence tools to screen thousands of resumes and applications within minutes.

However, concerns over the way AI processes data and finds patterns has led some in the tech and recruiting fields to worry about bias against jobseekers. 

“Right now, there are about one in four employers [26%] who are saying they are using AI” for some sort of HR purpose, said Clay Lord, director of programs for the SHRM Foundation, the charitable arm of the Society for Human Resource Management. “We anticipate that number is going to go up substantially quickly.”

Of the organizations that have been using AI, 64% say they use it for recruitment, interviewing, or hiring, to some capacity, according to a nationwide survey of 2,366 human resources professionals released in February by the SHRM.

In line with the sweeping popularity and interest of AI tools such as ChatGPT, which launched in 2022, more than 62% of professionals said their organization began using AI for HR functions sometime over the past year.

“Employers on average are spending about six and half seconds to assess a resume,” Lord said. “Right now, what that means is it’s a relatively unnuanced process.”

Hence, an increased appetite for AI tools that can speed up the process.

AI screening tools allow hiring managers to quickly sift through an increasing number of online applications and resumes, made larger by the fact that many employers now hire for remote positions.

“Now, they’re getting candidates from around the country, not just near the office,” said Stacie Haller, chief career adviser for ResumeBuilder.com. “That has tremendously increased the volume.”

Popular job search sites Indeed and LinkedIn recently launched a suite of AI tools for both employers and jobseekers.

Earlier this year, Indeed released Smart Sourcing to help employers identify and screen potential candidates on the business networking site as well as tools for jobseekers.

“Indeed’s AI-powered tools analyze the entire resume or application to create a comprehensive summary of a candidate’s qualifications,” Clint Carrens, career strategist for Indeed, said in an email. “This summary, which is shared with employers, provides a holistic view of the candidate, moving beyond the traditional keyword-based resume screening.”

LinkedIn, too, unveiled AI job search tools for jobseekers in June. Although the networking site doesn’t have applicant screening technology built into its LinkedIn Recruiter tool set for employers, it does use machine learning to help recruiters rank search results based on factors including work experience and skills.

Machine learning refers to the development of computer systems that can learn and analyze data using algorithms without explicit instructions.

“We firmly believe AI is here to assist with the hiring process — not displace it,” said Rohan Rajiv, director of product management at LinkedIn. “By helping with the more arduous tasks, it will help free up hirers’ time to focus on more strategic and human work, like speaking with and getting to know the candidates, and interviewing them for the skills they can bring to the job.”

Although generative AI tools like OpenAI’s ChatGPT have created a wave of intrigue among employers, tools that sort through online resumes and applications have been around for years in the form of applicant tracking systems, or ATS programs, recruiters said.

ATS software long has helped employers and recruiters track and organize job candidates through the screening and hiring process. Many of these systems provide keyword scanning tools that help hiring managers instantly sort resumes into distinct “yes” and “no” piles.

“A lot of the people that we deal with may not be as computer savvy,” said Jim Eddings, supervisor of Suffolk County’s One-Stop Employment Center. “We have to explain that the companies are using an ATS system to basically weed out or get resumes they deem good or match their algorithm.”

Eddings said the days of walking into a business and submitting a resume in person were long gone, and many job seekers — namely those who hadn’t been on the job hunt in decades — found the impersonal nature of trying to impress an algorithm frustrating.

“The way you look for work has changed a lot,” Eddings said. “You have to almost think differently.”

Adjusting to writing resumes to make it past screening software for the chance to maybe advance to the next level of the hiring process can be a headache for many jobseekers.

“This is the new way of the world,” Eddings said. “The resume was previously used to get the interview, and now it’s to get in the ‘yes’ pile to potentially get an interview. It’s another layer that’s been put in there. It is a very frustrating process.”

Local experts in the field of artificial intelligence said the use of such tools in the hiring process made sense, though concerns remained.

“The need is pretty clear,” said Steven Skiena, professor and director of Stony Brook University’s AI Institute. “For many jobs, businesses get hundreds or thousands of applicants.”

But Skiena said one of his main concerns was how strict AI could be in its filtering out candidates based on something potentially arbitrary like missing keywords.

“Say if you were a job seeker with a resume that didn’t have the magic words on it, and that for some reason the AI system doesn’t like your resume, you would get rejected by it every time,” Skiena said.

It’s possible, Skiena said, that a great candidate who otherwise might have some formatting irregularities in their resume or CV could be effectively locked out of the online application process without them knowing it.

“The advantage of human screeners, to a certain extent, is a degree of variability,” he said. “If there’s something in the screening algorithm that doesn’t like something weird about you or weird about your resume, you basically have no way of figuring that out.”

Additionally, Skiena pointed to examples of unintended bias being perpetuated by AI software with large companies such as Amazon.

In 2017, it was discovered by machine learning specialists at Amazon that a secret AI resume scanning tool the company was testing at the time was favoring male candidates. The bias was due in part to the data the specialists were using to train the AI; employee resumes over a 10-year period.

As a result, the AI observed that the bulk of resumes were from men — in line with broader gender disparities in tech — and likely concluded that male candidates were preferable as a result.

Despite concerns about unintended bias from AI in the screening process, the optimistic view is that the technology might help employers reduce bias in the hiring process by bypassing unconscious bias on the part of human screeners and hiring managers.

In that way, AI could play a role in helping employers expand diversity among its employees.

“In some sense, I would imagine that resume screening systems for classic examples of bias may be better than humans,” Skiena said

HiBob, an HR software company, is exploring the implementation of AI into its own line of products sometime next year. Company officials say they are taking their time in adding applicant screening tech.

“We are in the very early days of exploring AI capabilities and how we can use it effectively and responsibly to help our customers in leveraging it and becoming more efficient,” said Annie Rosencrans, director of people & culture at HiBob’s office in Manhattan.

“There are certainly a lot of unknowns, and we don’t want to be too quick on this,” she said.

Rosencrans said concerns of bias and the data used to train AI were concerns for HR professionals tasked with working with employees.

“That’s exactly the reason why we are actively pursuing AI, but not doing it too quickly,” Rosencrans said. “There’s a lot we don’t know yet. A big one is the bias piece. When we’re talking about people, it’s critical to get it right.

“We want to make sure we implement it in the most responsible and ethical way,” she added.

Haller, chief career adviser for ResumeBuilder.com, says she does worry about HR professionals becoming too reliant on new technologies in pursuit of efficiency.

“It has to find its place,” Haller said.

And while many organizations are looking at implementing AI to screen resumes, cover letters or applications, some are going as far as implementing some form of AI in the interview process.

A June 2023 survey from ResumeBuilder found that around 43% of employees involved in their company’s hiring processes said their organization planned to start using AI interview software this year, with around 10% having already implemented it.

Haller says she doesn’t imagine AI interviewers will take off as quickly as text screening tools.

“I don’t believe there’s a person on the planet that wants to be interviewed by a robot,” she said.