A research result showing that "It is better to leave company hiring personnel to algorithm" is released

BySamuel Mann

Humans have excellent instincts and intuition, but with regard to "employment", computers may have better insight.National Institute of Economic Research(NBER) examined "recruitment period" in human candidates and job seekers chosen by the algorithm, it became clear that the personnel selected by the algorithm worked longer.

Machines Are Better Than Humans at Hiring the Best Employees - Bloomberg Business
http://www.bloomberg.com/news/articles/2015-11-17/machines-are-better-than-humans-at-hiring-top-employees


NBER investigated the term of office of as many as 300,000 employees hired by 15 companies developing service businesses that do not require advanced skills such as call centers and data entry work. At the time of adoption, we will evaluate each job seeker using algorithms to recruitment staff of the company as "This person is an excellent personnel", but deciding which personnel to adopt is only the hiring person in charge of each company is.

In order to conduct this "personnel evaluation using algorithms", job seekers are asked to take a test. In the test, there are several questions to judge technical skills, personality, cognitive ability, whether it is suitable for occupation, etc. The answer of this test is analyzed by algorithm. In addition, the analysis results are categorized in such a way that "green" is a person with high potential, "yellow" is suitable person, "red" is a person with low evaluation. In analysisPrevious researchIt seems that the algorithm was built based on.

BySamuel Mann

As a result of the NBER survey, the group evaluated as "green" before recruitment had a longer overall employee term of 12 days than the group rated "yellow", and the group rated "yellow" was "red "It is clear that the overall average occupation period is 17 days longer than the group evaluated as" Algorithm "is proved to be operating properly. In the service industry that was the subject of survey, average employment period of employees seems to be very short and about three months. Associate Professor Michel Hoffmann, one of the survey members, commented, "This result is a big deal if we consider hiring tens of thousands." On the other hand, many employers in charge of recruitment adopted people who are not the best talents derived from algorithms, such as "conceit" and "prejudice".

Currently, "Employment Algorithm" seems to gradually become popular due to adoption cost and trouble of searching for a new employee, but "It is not said that highly skilled people with high evaluation of algorithms are productive" There is no pattern. Also, since this survey targeted "Service business that can be hired even if skill is not high," the ultimate "employment algorithm" that will select "excellent talent who will work long" will be completed It is likely to be ahead.