Researchers Develop System That Determines Political Leanings Based on Facial Features
A groundbreaking experiment supported by psychologists suggests that liberals and conservatives possess distinct facial features, a trait that artificial intelligence (AI) can detect with high accuracy.
This finding is now backed by a study asserting that combining artificial intelligence with facial recognition technology can determine someone's political orientation simply by analyzing their expressionless face.
The study, published in the American Psychologist journal, was conducted by researchers from the Stanford University School of Business. Before the experimental phase, all 591 participants completed a political questionnaire providing insights into their political beliefs.
Following this preparatory step, the AI was tasked with locating the participants along the political spectrum. The results showed the AI could accurately predict the political orientation of the individuals. The system's accuracy improved further when it was also informed of additional factors such as gender, age, and ethnic background of the subjects.
Before commencing the experiment, the researchers analyzed the facial feature differences between individuals identified as "most liberal" and "most conservative", incorporating these findings into their analysis.
The study revealed significant differences in the facial features of liberals versus conservatives.
These findings were then integrated into a database that categorized faces into different groups, which the artificial intelligence was successfully tested against.
While these revelations might seem innocuous and entertaining, they pose a somewhat concerning reality. According to the researchers, this sheds light on the capabilities of AI to elevate biometric surveillance to a new level, potentially making it more invasive. Particularly in the realm of political messaging, this technology could be exploited to target individuals based on their political orientations, inferred from as little as a photograph.