A team of researchers from Stanford University has developed an AI algorithm that after studying a single picture proves to be more than 70% accurate in estimating a person’s political affiliation. In his letter published in the journal Scientific reportThe group describes the construction and testing of its algorithm and how well it works.
Often, people wear hats, T-shirt Or buttons to announce them political affiliationEspecially during election cycles. But they can also declare their Connection According to the work done by the team in California, they are at least unaware of their facial expressions.
In this new effort, researchers built on prior work with an AI system they had developed that had better chance of estimating a person’s sexual orientation. This time around, he thought that if Facial expressions Or asan may say something about a person’s political beliefs – especially, if they were liberal or conservative.
The AI system that the researchers created was designed to learn new information from details that humans do not see. They then trained their AI system with data from a dating website and other sites, including user profile photos and political affiliations — the AI system looked for correlations of head posture with facial features and political orientation. He next tested his system with photographs and the system speculated on whether people identified as liberal or conservative. Algorithm Approximating between 71% of people who looked similar and 73% correct when guessing overall was corrected about 71% of the time.
Researchers were unable to tell what facial features their system correlated with political affiliation, but they did find some trends – head orientation and emotional expression, For example, appeared to provide some clues. For example, those who looked directly at the camera were considered more liberally oriented. On the other hand, those who show hatred are considered more conservative. Researchers acknowledge that it will be difficult for others to replicate their results correctly because they cannot share the photographic data they used for confidentiality reasons.
Mishal Kosinski. Facial recognition techniques can highlight political orientation from natural facial images, Scientific report (2021). DOI: 10.1038 / s41598-020-79310-1
© 2021 Science X Network
Quotes: More than 70% accurate AI algorithm on estimating an individual’s political orientation (2021, 14 January) https://naveenbharat.com/news/2021-01-ai-algorithm-accurate-person-political on 14 January 2021 Retrieved from .html.
This document is subject to copyright. No part may be reproduced without written permission, except for any fair which serves for the purpose of personal study or research. The content is provided for information purposes only.