Computers Judge Personality Better than Humans

Your Facebook likes allow computer models to judge your personality better than your coworkers, friends and family, according to study published in Proceeding of the National Academy of Sciences.

People leave digital footprints daily: from browsing history and online purchases to GPS locations and social media accounts.

According to the study, this allows machines to gauge human personalities without usual social-cognitive skills. In the study, the computer only needed 10 likes to beat a person’s work colleague, 70 likes to be more accurate than a person’s friend and 250 likes to beat a spouse.

That 250 number isn’t staggering either. According to the researchers, the average Facebook user likes around 227 pages. Facebook likes indicate a positive association with both online and offline entities, including products, activities, movies, books, sports, musicians, Websites and restaurants, among others.

Traits measured in the study and compared between the judgement of a human and a computer model included openness, conscientiousness, extraversion, agreeableness and neuroticism, referred to as the big five personality dimensions.

“Exploring the Likes most predictive of a given trait shows they represent activities, attitudes and preferences highly aligned with the big five theory,” the researchers write. “For example, participants with high openness to experience tend to like Salvador Dali, meditation or TED talks; participants with high extraversion tend to like partying, Snookie (reality shows star) or dancing.”

The human and computer-based judgements were compared against study participants’ self-ratings. However, the researchers write human judgement may be more accurate in subtle aspects of personality. During the study, human judges were limited to describing the human participant with a 10-item questionnaire.

Computer personality tools may assist in a variety of ways. Marketing messages could be tailored to a specific user, career recruiters may find better candidates for a job based on personality and scientists can collect personality data without surveys. “In the future, people might abandon their own psychological judgements and rely on computers when making important life decisions, such as choosing activities, career paths or event romantic partners,” the researchers write. “Such data-driven decisions (may) improve people’s lives.”

However, it’s a double-edged sword. The researchers suggest such information can also be used to manipulate and influence users.

“If I as a company want to sell you something, and I already know your psychological traits, I can produce a far more convincing marketing message to talk you into buying something that you don’t need,” said co-author Michael Kosinski to the Wall Street Journal. “Or think of the employees of a company working in places that aren’t so liberal. Local security forces, or even corporate competitors, could figure out which employees is gay, or which is atheist. This could impact the safety of employees.”  

Though users may get cagey towards digital technologies capable of inferring personality, the researchers hope technology developers and policy-makers will combat challenges by supporting privacy-protecting laws and technologies, and allowing users control over their digital footprint.