Machines Aren’t Better Than Humans At Picking Employees
I am unable to say this as emphatically as I once may have because new research suggests otherwise, but I still assert this is true. Two weeks ago I wrote a post about how the practice of blind interviews was misguided and basically contended that the opinions of hiring managers gained from in-person interviews were necessary to ensure cultural fit and employee retention. In contrast, a new study by the National Bureau of Economic Research reveals that computers do a better job at hiring employees and could further validate the idea that the hiring process does not need the hands on evaluation of a hiring manager.
The researchers looked at fifteen companies and more than 300,000 hires in low-skill service sector jobs. Tenures were compared of employees who had been hired by a human with those who had completed a job test and were picked by an algorithm. The test evaluated candidates for technical skills, cognitive skills, personality and job fit. The results showed that candidates picked by the algorithm remained longer with the company and were no less productive than those chosen by recruiters.
Based on this research one article concluded with its title: “Machines Are Better Than Humans at Hiring The Best Employees.” Is the case open and shut though? Now the computer’s algorithm was solid. Its top picks, labeled as green, stayed 12 days longer than those it labeled as yellow and yellows stayed 17 days longer than the reds. And yes, when a recruiter, based on their gut, deviated from the algorithm’s suggested green pick and hired a yellow, the yellow pick left earlier than the green picks. Given all of this I still can’t affirm that a human is worse or unnecessary in the hiring process. The employees evaluated in this study filled lower level jobs such as you would find in a call center and the average tenure of these employees was only three months. We cannot conclude that machines are better based on a comparison between two processes hiring entry level employees. Is the human hiring process for a minimum wage earner who likely won’t work more than ninety days equally as sophisticated as the hiring process for someone making $80K a year? To claim machines are better, the same comparison must be made of the hiring processes at different salary levels.
I’m also curious to know how many green candidates dropped out of the hiring process rather than submit to a rigorous evaluation that tested their skills, personality, etc. for a low paying job. Based on their gut, how many green candidates did recruiters successfully identify and in less time? With emerging technologies such as video interviewing, recruiters can see/hear candidates in minutes and make informed hiring decisions. How do both hiring processes compare in terms of time spent placing the employee and is this speed in filling the position taken into account as a positive?
What becomes of diversity in a computer based hiring process? Presumably age, ethnicity and gender data is not crunched by the computer. While this may sound like a measure designed to eliminate discrimination a blind process does not necessarily promote diversity. How many minorities might be at a skills or cultural disadvantage because their socio economic status did not provide them with the same educational and social resources their white peers were afforded? Would these individuals possibly be eliminated by a computer but hired by a human?
Lastly, though the research suggests the computer with its battery of tests is better than a human, let us not forget who designed the computer and the battery of tests in the first place. Ultimately humans created and programmed a better hiring process, at least at the lower levels, so in that respect we should not yet consider ourselves obsolete.