An innovation from Stanford University researchers could determine the political affiliation of people. According to Scientific Reports, the researchers’ team curated an Artificial Intelligence algorithm, which has an accuracy of 70-72% at deducing a person’s political orientation. In the journal, the scientists have provided all the details of developing and testing the algorithm.
Suppose the technology behind the facial recognition would help greatly in times of election. However, the way the facial recognition AI came into being might be a shocker. Before curating this cutting-edge AI algorithm, the scientist developed a facial recognition system which could guess people’s sexual orientation. However, the latter is better, robust, reliable, and provides quite accurate results.
Michael Kosinski, psychologist, Stanford University, revealed that the team had to train the AI algorithm using images of over 1 million USA and UK citizens. To make the AI better than the prior experiment, they also took people’s age, gender, & political views into consideration. The team didn’t breach any people’s privacy while making the AI learn about the political affiliation; rather, all pictures were acquired from their Facebook profile and virtual dating platform.
The scientists completely rely on physiognomy to determine people’s preferred political personality or party. However, the research is yet to determine the cause of how AI determines and correlates people’s affiliation. According to the study, the political affiliation was determined from the emotional expression and head orientation of their pictures. The researchers have expressed that they cannot share this photographic data as it will breach people’s privacy.