Viewpoint: AI Says I Have a 12% Chance of Succeeding at My Job
Neurological pattern matching vets job applicants, but how well?
According to artificial intelligence (AI), I have only a 12 percent chance of making it one year as a writer. Having been a writer for the last 25, I'm not too concerned.
If you're applying for work at Unilever or Accenture, though, you might want to be. Frida Polli, Ph.D., CEO and cofounder of Pymetrics, says it and other companies use Pymetrics' online, neuroscientific AI tests to predict applicants' job success.
Pymetrics offers 12 cognitive exercises; they are the same whether you as an individual are testing your career skills or an employer is looking for applicants that would be good matches for job needs and company culture. One of the exercises is a round of solitaire. Two ask how much money you'd give a stranger, knowing you might not get it back. In another exercise, you click arrows as images flash on a screen: Right if the picture's black or blue, left if it's red. Pymetrics compares your results to a control group of successful existing employees, and voila! A strong match means you're one step closer to the job.
But if your neurological processes and those of existing staff don't sync, it's bad news. "[O]nce we go in and look at high-performing individuals … different traits come to the surface for different roles at different companies," Polli said. The system uses your test performance to generate a list of career matches ranked by percentage of how likely you are to remain in each position for more than a year. The fewer trait matches, the lower your percentile chance.
According to Polli, each corporation uses results differently. Accenture marketing manager Sam Hyland says the company uses Pymetrics' predictive modeling in multiple ways, but he declined to elaborate. Unilever did not respond to a request for interview, but Polli says the company has used Pymetrics as a campus-level recruitment screening in the past two years. "We're doing a fit match," Polli explained. "We're telling someone, both the individual and the company, 'Hey, the fit is strong,' or 'The fit is medium,' or 'The fit is weak.' "
Pymetrics also says it aids in diverse hiring by removing race and gender identifiers from the selection process, providing analysis based strictly on candidates' brains.
But predictive modeling helped create corporate America's diversity problem. Certain companies limited their definition of success to existing employees, then—consciously or subconsciously—restricted job offers to those who looked just like them. I didn't test like the writers from Pymetrics' control group, but I have been one my entire adult life, despite that 12 percent match.
With questions about the reliability of test results and possible obstacles to diverse hiring, why would companies use AI to screen candidates? Because employee turnover costs businesses so much money, according to Danny Nelms, president of workforce research company Work Institute. Job attrition at U.S. companies adds up to approximately $536 billion per year, he said. According to the Bureau of Labor Statistics' Job Openings and Labor Turnover Survey (JOLTS Report), more Americans left their jobs in October 2017 than have since 2010, with 3.2 million employees quitting that month alone. Nelms says as the economy continues to grow, workers have more options—which translates into more mobility. It's a matter of when—not if—new hires will quit, and it's easy to understand why businesses would invest in tech that promises to predict pre-hire when someone will go. "We have strong correlation with reduced attrition," Polli said, explaining that current clients experience a 14 percent to 60 percent increase in retaining new hires past the first year.
In this data-driven scramble to screen out the short-timers, how many great hires do corporations miss? Unless I'm applying to be a casino dealer, should my card-playing skills keep me from getting a job? Other factors are often at play when employees decide to leave an employer. Nelms said 75 percent of turnover stems from something the company—and not the individual—controls. For example, I quit my first journalism job because it paid only $6 an hour. At another, I was sexually harassed by a co-worker who followed me home at night. If the goal is preventing turnover, my neurological makeup may not be the right place for some employers to start.
"I don't know for sure that we will be able to [prevent turnover] with AI," Nelms said. "One thing that it's almost impossible to predict is how this person is going to interact with the workplace conditions." Plus, "workplace conditions are in a constant state of flux."
Polli agrees that work environments can be highly individualized: "Because the methodology that we use is very precise, there are certain things that do surface across the same role at different companies. But there's actually a lot of variability as well, and so that's why we strongly believe that it's better to go in and do a custom profile per role per company rather than assuming that, hey, if you're looking for—I don't know—a person in marketing, these are the three things you should look for."
Terena Bell is a freelance writer in New York City.
Was this article useful? SHRM offers thousands of tools, templates and other exclusive member benefits, including compliance updates, sample policies, HR expert advice, education discounts, a growing online member community and much more. Join/Renew Now and let SHRM help you work smarter.
Advertisement
An organization run by AI is not a futuristic concept. Such technology is already a part of many workplaces and will continue to shape the labor market and HR. Here's how employers and employees can successfully manage generative AI and other AI-powered systems.
Advertisement