A Russian research team has developed a neural network that can speculate on one’s traits of character by analyzing the person’s photo—and do it better than humans.

The new artificial intelligence that came from scientists at the Higher School of Economics (HSE), the Open University of Humanities and Economics, and a Russo-British start-up called BestFitMe is believed to be more efficient in character prognostication by photo than the average human can achieve, and more accurate than the current competition from among machine learning solutions.

The neural network has studied tens of thousands of selfies and learned to evaluate at least five parameters: honesty, neuroticism, extraversion, conformity, and openness. Honesty proved to be the best-guessed trait of these all.

Curiously enough, the algorithm has shown higher efficacy when dealing with female photos, compared to male ones.

A machine learning method like this one could help step up the assessment of candidates in situations where swiftness is key or an evaluator has to work with huge pics databases. Markets for the technology could primarily include headhunting agencies, dating websites/apps and customer service, all interested in selecting most friendly and communicative faces to deal with.