Conservative female politicians tend to appear happier and more attractive in pictures than liberal politicos, a new artificial intelligence study conducted in Denmark found.
Published in the Nature-owned journal Scientific Reports in March, the research found AI can predict a person’s political ideology with 61% accuracy by analyzing just one headshot.
The scientists inputted about 3,200 publicly submitted photos of political candidates who ran in the 2017 Danish municipal election into Microsoft Azure’s Face API tool to assess the person’s emotional state.
The analysis found 80% of the faces displayed a happy expression, while 19% read as neutral.
“For females (though not males), high attractiveness scores were found among those the model identified as likely to be conservative,” read the findings. “These results are credible given that previous research using human raters has also highlighted a link between attractiveness and conservatism.”
The results were even more accurate for men, 65%, before the researchers stripped their photos of visual imagery other than the man’s face — such as shirt collars.
Left-leaning male politicians showed more neutral, less happy faces than their conservative counterparts, the study found.
“Attractiveness was not the only correlate of model-predicted ideology,” the scientists explained. “We also found that expressing happiness is associated with conservatism for both genders.
“Previous work has found smiling in photographs to be a valid indicator of extraversion,” they continued. “And while extraversion is not broadly associated with ideology some studies have found that right-wing politicians are more extraverted.”
Scientists noted that “because attractiveness generally helps electoral success, all candidates are incentivized to provide an attractive photograph.”
“Politicians on the left and right may have different incentives for smiling — for example, smiling faces have been found to look more attractive which is comparatively important for conservative politicians,” the paper reads.
“Future work is needed to explore the extent to which happy faces are indicative of conservatism outside of samples of politicians.”
Of greater concern is the “threat to privacy posed by deep learning approaches” using publicly available data.
This is not the first time AI has raised warning flags.
In March, Facebook removed AI-generated deepfake sexual social media ads that used the likenesses of actresses Scarlett Johansson and Emma Watson.
And a recent report found generative AI will cause significant disruption in jobs held by “higher-wage knowledge workers” whose roles were “previously considered to be relatively immune to automation.”