In the United States, scientists at Stanford University have created an algorithm that, with an accuracy of 72%, can determine a person’s political orientation from a photo.
To train the algorithm, a psychologist from Stanford University (USA), Michael Kosinski, used images of more than a million people from the USA, Canada, and the UK, as well as data on their gender, age, and political views. The pictures were taken from Facebook profiles of people, as well as from an online dating service.
Based on the images, 2000 features were identified that distinguish people by their political orientation. However, the algorithm has learned to single out only liberals and conservatives.
The specified approach to the method used in physiognomy
Now the algorithm works only with residents of the USA, Canada, and the UK. In the future, the author expects to use it in other countries.
Earlier, Kosinski created a system that uses similar algorithms to determine a person’s sexual orientation.