This Software Could Be Better At Hiring Than Any Human

Call it HR: Rise of the Machines

Business Insider

Computer
programs can draw conclusions about hirability based on your digital
footprint, which includes data like Facebook "likes."

By Jacquelyn Smith



What you "like" on Facebook may seem inconsequential now - but there's a good chance that will change in the future.



A new study, cited by The New York Times, finds that computer models can
draw accurate and detailed conclusions about your personality and
creditworthiness, among other things, based on your Facebook "likes."



    

According to researchers Youyou Wu and Dr. David Stillwell of the
University of Cambridge, and Dr. Michal Kosinski of Stanford University,
employers may eventually use this technology to make important hiring
decisions.



In a paper on the study, the researchers write: "Although accurate
personality judgments stem from social-cognitive skills, developments in
machine learning show that computer models can also make valid
judgments." They found that these models may even make better judgments
than humans.



Using a sample of 86,220 volunteers who completed a 100-item personality
questionnaire, the researchers determined that computer predictions
based on a generic digital footprint (Facebook "likes") are more
accurate than those made by the participants' Facebook friends using a
personality questionnaire.



The researchers found that someone who "likes" Nike and In-N-Out Burger, for example, is likely a calm and relaxed person.



"Computers outpacing humans in personality judgment presents significant
opportunities and challenges in the areas of psychological assessment,
marketing, and privacy," they write in the paper.



Another process this may have an affect on? Hiring.



"Currently, occupational psychologists evaluate people's characteristics
and decide the fit between people and jobs," Wu tells Business Insider.
"It's very likely that in the future this process of assessing
personalities and determining how someone's characteristics are related
to a certain job will be automated using computer models like ours."



Stillwell says there are many benefits to using computer models like the
ones he and his colleagues have created - which are only being used for
their research projects for now - as long as they are implemented with a
respect of privacy and ethics. "One, computer models are cheaper than
human capital; two, computer models are more efficient and can be
applied on a large scale; and three, they generate more reliable
results, as computers can use big data to detect unobservable patterns
between likes and personality, or between personality and jobs."



Eventually, he says, employers will be presented with a list of job
candidates that computers deem the best matches, without knowing why
they are suitable. "Besides the benefits we already mentioned, this
approach would help promote equality in the selection process and avoid
human biases prevalent in occupational settings," Stillwell adds.
"Computers do not favor people of certain gender, race, or personality."



But of course there is some apprehension.



"I think people, from a job candidate's perspective, might be at first
worried about not being able to present themselves in the way they want
anymore," Kosinski says.



But they needn't be worried, he argues, since if candidates present
themselves in an inaccurate way on social media, it could eventually
lead to a mismatch between their characteristics and the job.



As for whether employers will begin using these models for hiring purposes and when, the researchers are unsure.



"It takes time for companies to switch gears and accept a new recruiting
method," says Kosinski. "I think it's likely that some companies will
experiment with computer's evaluations, and use it as a reference in
addition to other traditional metrics. There are also legal and ethical
concerns that need to be addressed before any implementation," he
explains.



"For instance, users need to understand which of their personal data is
out there, how it is being used, and how it might be used," he
continues. "We also need to enable users to take full control of their
data and decide for which purpose it is to be used. Both aspects have
relatively well understood technological solutions, but their
implementation may require user awareness and, perhaps, some nudges from
policy makers."



Wu says any companies that collect data on individuals, like Facebook
and major banks, should take it upon themselves to inform the public
about how that data can be used to benefit the users themselves. "I
believe that if users have a better sense of how their online
experiences can be improved by letting their data be analyzed, and they
have the control over how and what data are analyzed, they will be more
motivated to share their data."