The Reason Job Applicants May Withdraw When Informed of AI Screening

Job applicants are willing to acknowledge some degree of objectivity in AI-driven hiring algorithms when used at early stages of the process, like screening resumes. 

But they are less likely to proceed with an application if they are informed that AI will be conducting an interview or is otherwise completely responsible for the employment decision. The reason is that they expect some 'right' to promote themselves and learn more about the employer. JL 

Lisa Ward reports in the Wall Street Journal:

Researchers found in certain instances, like the screening of applications, study participants usually accepted some degree of automation. But in other instances, like interviews, automation could deter job seekers from applying for a position. Participants who were told the hiring would be fully automated tended to believe they had less agency or voice in the final outcome. The study results suggest this concern outweighs appreciation of AI’s lack of bias. During interviewing, applicants expect personal interaction, to give them an opportunity to sell themselves and to learn more about the company.

If job seekers knew companies were using artificial intelligence to fill open positions, would it stop them from applying for the job?

The answer, according to a recent study, is yes—sometimes.

The researchers found that in certain instances, like the screening of applications, study participants usually accepted some degree of automation. But in other instances, like interviews, the study suggests, automation could deter job seekers from applying for a position.

Companies contending with recent labor shortages are increasingly turning to AI as a way to facilitate and speed up the hiring process. AI can be used in such tasks as screening job applicants for basic qualifications, checking for professional credentials and licenses, evaluating video statements, interviewing candidates and conducting competency assessments.

---

The new research underscores when using AI in hiring could be counterproductive. For instance, in one part of the study participants were shown fictional job postings and then asked if they intended to apply for the position. The researchers found that if the job posting said AI was used to both screen applicants and conduct interviews, participants’ intention to apply to the position averaged 2.77 on a six-point scale, with 6 reflecting the highest intention to apply. If AI was used only for the screening process, participants’ intention to apply averaged 3.73.

In another experiment, the authors also found that study participants saw pros as well as cons in the use of AI in interviewing.

Participants who saw a job posting stating that AI was used to both screen applicants and conduct interviews expected the hiring process to be more consistent in its judgments than those who saw postings with less AI involvement, ranking the process at an average of 3.66 for consistency on a five-point scale, with 5 being most consistent. Participants where AI was to be used to screen applicants but not interview them ranked the process at 3.48 for consistency, and participants where the posting made no mention of automation ranked the process at 3.16 for consistency.

On the other hand, participants who were told the hiring would be fully automated tended to believe more than others that they had less agency or voice in the final outcome. Overall, the study results suggest this concern tends to outweigh the appreciation of AI’s lack of bias at the interviewing stage.

“A hybrid approach where companies use AI in some tasks but not others may be a way to get the best of both worlds,” says Jenny Wesche, a co-author of the study and a postdoctoral research fellow at the Free University of Berlin. Participants may have been more open to automation earlier in the hiring process because they had little expectation of direct interaction at that stage and could see some benefits from using AI—such as less-biased decisions and the avoidance of problems like nepotism, she says. But during the later stages, she suggests, applicants expect personal interaction, to give them an opportunity to sell themselves and to learn more about the company.

“AI is not inherently good or bad,” Dr. Wesche says. “It just very much depends on the context it’s used.”