Jan. 27, 2021 -- A facial recognition algorithm was able to correctly identify conservative versus liberal orientation remarkably better than chance, human accuracy, or a personality questionnaire, according to an article published in Scientific Reports.
A study used social media photos of 1,085,795 participants – including 347,000 people of color, from the United States, United Kingdom and Canada and their self-declared political orientation, age, and gender, to identify political inclination.
The analyzed photos combined many potential cues to political orientation, ranging from facial expression and self-presentation to facial morphology, the researcher says.
The open-source facial recognition algorithm extracted more than 2,000 data points for the study, which were then cross-validated to compare the images with liberal and conservative faces. According to the results presented in the article, the algorithm was able to correctly predict the political orientation in 72% of cases in the largest dataset used. The accuracy varied between 70-73% across all datasets before and 65-71% after controlling for demographics.
The author notes that this analysis does not imply that liberals and conservatives have innately different faces, but rather that other transient features account for the predictability of political orientation. These findings might have a tremendous impact when applied to large populations in high-stakes contexts, such as elections, he concludes.
Comments
Post a Comment