AI capable of determining a person’s political affiliation based on their photo, liberals find the camera in the face while conservatives have an aversion
- Stanford experts build an AI that could guess political affiliation through a photo
- It is trained with more than one million images of appointments and Facebook
- The AI focused on head orientation and facial expressions during guesswork
- It found that most liberals look at the camera while conservatives look disgusting
The Stanford research that made headlines in 2017 about designing an AI that uses ‘viewpoints’ to determine a person’s sexual preference is back to another controversial system.
Dr. Michal Kosinski claims that he has a face recognition algorithm that can identify whether someone is liberal or conservative, based on a single photo – and with more than 70 percent accuracy.
The technology, which builds on the 2017 AI, has been trained with more than a million images from dating sites and Facebook and programmed to focus on expressions and attitudes.
Although Kosinski and his team could not identify the algorithm associated with a political preference, they did find some trends such as co-orientation and emotional expression in pictures.
Some examples include that people who looked directly at the camera were labeled as liberal and that those who were disgusting were considered more conservative.
Scroll down for video
The technology has been trained with over a million images from dating sites and Facebook and programmed to focus on expressions and attitudes. The machine learning system cuts the size of the face and changes it to reduce the capture of non-facial features
The study, published in Nature, says that when people are asked to distinguish between two faces – one conservative and one liberal – it is correct about 55 percent of the time.
“Because people lack some clues or misinterpret, their low accuracy does not necessarily represent the limit of what algorithms can achieve,” the study reads.
‘Algorithms perform in recognizing patterns in large data sets that no human can ever process, and we increasingly perform in visual tasks, ranging from the diagnosis of skin cancer to facial recognition to face-based judgments about intimate traits, such as sexual orientation (76% vs. 56%) 7, personality (64% versus 57%; derived from Pearson’s), and – as shown here – political orientation. ‘
Researchers used a sample of 1,085,795 participants from the U.S., Canada, and the United Kingdom, along with their self-reported political orientation, age, and gender.
The Stanford study that made headlines in 2017 for designing an AI that uses ‘viewpoints’ to determine a person’s sexual preference (photo) is back with another controversial system.
The study noted that the ethnic diversity of the same included more than 347,000 non-white participants.
The machine learning system cuts and resizes the face to reduce the capture of non-facial features.
In terms of identifying U.S. images, the AI was 72 percent accurate.
Similar accuracy was seen in the Canada sample, 71 per cent, and the UK with 70 per cent.
Researchers used a sample of 1,085,795 participants from the U.S., Canada, and the United Kingdom, along with their self-reported political orientation, age, and gender. In terms of identifying U.S. images, the AI was 72% accurate. Similar accuracy was seen in the Canada sample, 71%, and the UK with 70%
The highest predictive power was followed by head-straightening (58 percent), followed by emotional expression (57 percent).
Liberals tended to look more directly at the camera, were more likely to express surprise and less likely to express disgust – those with a look of disgust were branded as conservative.
‘In other words, a single face image reveals more about a person’s political orientation than the answers to a fairly long personality questionnaire, including many items that seem to be related to political orientation (eg’ I treat all people equally ‘ or ‘I believe it’s also a lot of tax money going to support artists’), reads the study.