Mindclones from Social Media: New Research from Stanford Suggests Feasibility
In an episode of the popular dark sci-fi show ‘Black Mirror‘, realistic digital personalities of the dead are recreated from data alone. London-based firm “Lean Mean Fighting Machine” has developed an artificial intelligence system that can analyse a person’s Twitter feed, and then impersonate them after death. But just how realistic is this idea?
This new study, published Jan. 12 and conducted jointly by researchers at Stanford University and the University of Cambridge, compares the ability of computers and humans to make accurate judgments about personalities. People’s judgments are on their familiarity with the judged individual, while the computer uses digital signals derived from Facebook “likes.”
The researchers were Michal Kosinski, co-lead author and a postdoctoral fellow at Stanford’s Department of Computer Science; Wu Youyou, co-lead author and a doctoral student at the University of Cambridge; and David Stillwell, a researcher at the University of Cambridge. By mining a person’s Facebook “likes,” a computer program was able to predict a person’s personality more accurately than most of their friends and family. Only a person’s spouse came close to matching the machine’s performance.
The software’s predictions were based on which articles, videos, artists and other items the person had liked on Facebook and predictions matched the subject’s own scores on the five basic personality dimensions: openness, conscientiousness, extraversion, agreeableness and neuroticism.
Kosinski, a computational social scientist, pointed out that “the findings also suggest that in the future, computers could be able to infer our psychological traits and react accordingly, leading to the emergence of emotionally intelligent and socially skilled machines.”
“In this context,” he added, “the human-computer interactions depicted in science fiction films such as Her seem not to be beyond our reach.”
He also said the research advances previous work from the University of Cambridge in 2013 that showed that a variety of psychological and demographic characteristics could be “predicted with startling accuracy” through Facebook likes.
In the study, personality self-ratings of 86,220 volunteers were collected using a standard, 100-item long personality questionnaire. Human judges, including Facebook friends and family members, expressed their judgment of a subject’s personality using a 10-item questionnaire. The results showed that a computer could more accurately predict the subject’s personality than a work colleague by analyzing just 10 likes; more than a friend or a roommate with 70; a family member with 150; and a spouse with 300 likes.