A recent study published in the journal American Psychologist reveals that the evolution of facial recognition combined with AI (artificial intelligence) can analyze with impressive accuracy a person's political beliefs by analyzing their detailed and unconscious facial expressions.
The study was conducted by experts at Stanford University's Graduate School of Business. According to the researchers, before the experiment, 591 participants were asked to fill out a political questionnaire, thus revealing their political beliefs. The same participants were then analyzed by an AI algorithm developed by the researchers, which attempted to determine their place on the political spectrum.
Read more: Jennifer Lopez's film teaches us to appreciate AI
The researchers found that an algorithm can accurately predict a person’s political orientation, even when the information associated with the person is limited to age, gender, and ethnicity. The algorithm’s predictive power is further enhanced by access to these demographic details. The accuracy of this tool was analyzed under various conditions and compared to the accuracy of job interviews in predicting career success or the extent to which alcohol can lead to aggression.
Before conducting this experiment, the researchers investigated the differences in average facial features between “more liberal” and “more conservative” men and women, and the findings of that study were incorporated into their analysis. The study reveals that there are significant differences in facial morphology between liberals and conservatives. Specifically, liberals have “smaller faces,” “down-sloping lips and noses,” and “smaller chins” compared to conservatives, according to the researchers.
According to this theory, facial dimensions can reveal political preferences: a thin face is often associated with progressive views, while a wide face may indicate support for Donald Trump. This analysis is based on the idea that the face and head shape destiny, suggesting that social perceptions of appearance can influence development . Researchers support this theory by showing how social expectations about appearance can shape a person's character and beliefs.
After the researchers were able to construct a hypothetical association between specific facial features and political beliefs, they created a database of faces that fit those categories. They then tested a algorithm to see if it could accurately predict political leanings based on facial appearance. According to their findings, the method proved effective.
"We have demonstrated that an individual's political orientation can be accurately predicted by both humans and algorithms, based on neutral facial photographs, regardless of parameters such as age, gender, and ethnicity. These findings reveal an interesting connection between political preferences and inherent, physical facial features, which are largely independent of personal choice," according to the study.
See also: Artificial intelligence faces diversity problems
The researchers stressed: “Most importantly, our findings reveal that widespread biometric surveillance applications pose a greater threat than we had initially estimated.” This is particularly worrying in cases where they are used to target political messages online, as the ability of artificial intelligence to determine an individual’s political orientation could prove particularly important.
Source: gizmodo
