A team of Russian researchers has developed an artificial neural network that can guess character traits from photographs. Curiously, the algorithm is better for women than for men. The authors intend to offer technology to dating sites, training video producers and customer services.
Since ancient times, physiognomists have tried to connect a person's appearance to his character, but most of their ideas did not withstand rigorous scientific verification. Studies in which participants are asked to "rate" someone's personality from a photograph usually always give very mixed results. Why should it make a difference with a selfie?
However, there are several theories (indeed questionable for me) on the fact that the essential characteristics of communication can be determined by carefully studying the face of the interlocutor. The results of the most recent neurobiological studies show that, instead of paying attention to individual characteristics, the brain processes facial images entirely and on a holistic basis.
Tell me what selfie you take and I'll tell you who you are
Scientists from HSE and the Open University of Humanities and Economics, together with the Russian-British startup BestFitMe, they developed a group of neural networks to make reliable predictions about the nature of face photographs. The effectiveness of artificial intelligence thus educated turned out to be superior to that of the average person and even that of similar machine learning systems. The ability to "guess" the character of a person from his photo was 58%, not 50% like up until now.
After studying 31.000 selfies, the neural network was able to evaluate five parameters: honesty, nervousness, extroversion, conformity and openness. Honesty is the parameter that AI has read better than others. And the conclusions drawn from the study of female photographs turned out to be more accurate than assessments on men's faces.
The method can become a support for the evaluation of candidates in situations where speed is needed or it is necessary to manage a large volume of photos. The technology can be used by recruiting agencies, dating sites and applications.