Hiring managers may soon be using artificial intelligence to determine what type of person you are, and whether you’re a fit for the job.
Researchers from Moscow’s HSE University and Open University for the Humanities and Economics have demonstrated that artificial intelligence may be better able to assess personality from selfies than humans. The researchers have teamed up with a Russian-British business start-up BestFitMe to train artificial neural networks to make reliable personality judgments based on photographs of human faces.
12 thousand faces (minus celebrities and cats)
To test the AI they conducted a study with a sample of 12 thousand volunteers who completed a self-report questionnaire measuring personality traits based on the “Big Five” model. The “Big Five” traits are conscientiousness, neuroticism, extraversion, agreeableness, and openness, as defined by contemporary personality psychologists who suggest that there are five basic dimensions of personality.
According to a research brief, volunteers uploaded a total of 31 thousand selfies online which were randomly split into a training and a test group. The pictures were preprocessed to ensure consistent quality and characteristics, and to exclude faces with emotional expressions, as well as pictures of celebrities and cats. An artificial neural network was trained to break down each image into 128 invariant features and use these features to assess personality traits.
The AI performed better than chance
The performance of the AI was reportedly able to make above-chance judgments about conscientiousness, neuroticism, extraversion, agreeableness, and openness based on the selfies, and the personality judgments were consistent across different photographs of the same individuals.
The AI can correctly guess the relative standing of two randomly chosen individuals on a personality dimension in 58% of cases as opposed to the 50% expected by chance. This indicates that an artificial neural network relying on static facial images outperforms an average human rater meeting someone in person for the first time. Conscientiousness was the most easily recognizable trait of the five and personality predictions based on female faces appeared to be more reliable than those for male faces.
The researchers believe there are a number of potential applications for this technology, such as proposing products that are the best fit for the customer’s personality or selecting the possible best matches for people in dyadic interactions, like customer service, dating, or online tutoring. There may also be potential application for use when assessing candidates for jobs. While it’s not common practise, or advisable, to have a picture in your resume, it’s important to have one on LinkedIn.
While a 58% accuracy rate isn’t a stunning accuracy rate, it’s eight per cent higher than chance, and when combined with other forms of information, this it could be another tool in the kit used to find the right candidate. Also, the accuracy may improve over time.