AI could identify political ideology such as 'conservative' or 'liberal' just by looking at a face
In order to determine tendencies such as 'this person is a conservative' or 'this person supports the liberal party,' it is usually necessary to analyze a person's words and actions. However, a research team at the Stanford Graduate School of Business claims that by combining facial recognition technology and AI, it is possible to accurately assess a person's political orientation just by looking at their face.
Facial Recognition Technology and Human Raters Can Predict Political Orientation from Images of Expressionless Faces Even When Controlling for Demographics and Self-Presentation
AI Can Tell Your Political Affiliation Just by Looking at Your Face, Researchers Find
AI can predict political orientations from blank faces as researchers fear 'serious' privacy challenges
https://nypost.com/2024/04/23/tech/ai-can-predict-political-orientations-from-blank-faces-as-researchers-fear-serious-privacy-challenges/
Michael Kosinski, an organizational behavior scientist at the Stanford Graduate School of Business, and his colleagues first administered a political questionnaire to 591 subjects to ascertain their political beliefs.
The subjects' faces were then scanned using a proprietary AI algorithm, and the resulting image data was used to assess their political ideology. The results were then compared with the subjects' responses to a questionnaire survey. The results of the experiment showed that the system was able to accurately determine the subjects' political ideology from just the facial images, even without data such as the subjects' age, gender, or cultural background.
During the experiment, subjects were required to remove all makeup, wear black T-shirts, and wear hair ties, hairpins, and headbands, and hide their clothes with binder clips. They were also required to remove all jewelry and remove facial hair if necessary. In other words, the AI did not determine the subjects' political orientation based on factors such as clothing, accessories, or makeup.
The accuracy of the algorithm's predictions is 'similar to predicting the likelihood that a job interview will lead to job success, or the likelihood that alcohol will provoke aggression,' the team said.
In addition, this experiment showed that 'liberals tend to have smaller lower faces, smaller chins, and smaller overall faces than conservatives.' In other words, this experiment showed that there is a tendency for 'big faces are conservatives' and 'small faces are liberals.'
'Facial appearance can lead to subsequent psychological traits. In this study, we were able to identify participants' political ideology by scanning their faces,' the research team said.
The research team is currently working to create a database of ideologically-based faces based on the results of this study, and is working to further improve the accuracy of the prediction. 'We have demonstrated that our algorithm can predict political orientation from facial images, regardless of factors such as age, gender, or ethnicity,' the research team said.
However, the research team pointed out, 'This technology is beyond an individual's control.' 'Our findings suggest that pervasive biometric surveillance technologies are more surprising than previously thought. The ability of AI to infer a person's political leanings could be crucial, particularly when it comes to targeting online political messages.'
'The algorithms used in this case can easily be installed on mobile phones and still capture a significant amount of information,' Kosinski said. 'Many people are unaware of the information they are revealing by posting photos online.'
Related Posts:
in Software, Posted by log1r_ut