An algorithm for determining political orientation from a photo for a face recognition system was compiled by a psychologist from Stanford University (USA) Michael Kosinski.
According to the latest research, published in the scientific journal Nature, facial recognition can help determine a person's political views from a photograph.
Psychologist at Stanford University (USA) Michael Kosinski has developed a facial recognition program, the algorithm of which is supposedly able to determine from a photograph what political views a person holds.
In order for the program to train the algorithm as much as possible, the specialist used photographs of more than a million people from the USA, Canada and the UK, as well as data on gender, age and political views. Moreover, about 350 thousand people were not representatives of the Caucasian race.
The algorithm revealed about two thousand signs by which it is possible to determine whether a conservative or a liberal is captured in the photo. The author of the article claims that the neural network guessed correctly in 72% of cases, while volunteers - in 55%.
Currently, the algorithm works only with residents of the United States and Canada, as well as the UK, but in the near future Kosinski expects to slightly expand the boundaries of application.
It is worth noting that several years ago Kosinski created a system that recognizes a person's orientation from a photo. But many criticized this invention, because in some countries homophobes are allowed at the legislative level.