A study by a group of scientists from the University of Cambridge, which included 50 people, led to the creation of an algorithm that provides accurate information about the religion, race, gender and political affiliation of users, based solely on Facebook likes.
The algorithm was 88% accurate in predicting men, 95% for race, 85% accurate in differentiating between Republicans and Democrats, and even in 82% of cases it gave the correct result regarding belonging to the Christian or Islamic religion.
When it comes to various character traits, accuracy was determined between 62 and 75% of cases.
Problematic from the point of view of human rights
According to the BBC, this study could be extremely important for companies when it comes to personalized marketing.
Nevertheless, NBC journalist Mišel Kozinski warns that this can also represent a threat to human freedoms: "I appreciate automatic book suggestions, or Facebook's recommendations of important things, but it is also possible to imagine situations in which the same data and technology are used to predict political views or sexual orientation".
Bonus video:
