A group of researchers from Stanford University has recently developed a study where they tested an AI network for its accuracy on identifying a person’s sexual orientation just by analyzing photos. The technology looked at photos posted on a dating website and, then, based only on this sample, could tell if a person was gay or not. Of course, this study sparked a lot of controversy.
Gay people seem to exhibit different physical traits than straight people
The results of the study showed how the algorithm had an accuracy of 81 percent when it came to men, and 74 percent for women. It made use of a sample of 35,000 close-ups of people who set their photos as public on a dating website.
Researchers were able to perform the task with the help of deep neural networks, which can take different details from photos and connect them to groups they might belong to. Those men identified as gay exhibited more feminine traits in their style of grooming, or in their expressions. Also, their nose was a little longer, their jaws narrower, and had a bigger forehead.
The study has faced a lot of controversy
Since certain physical traits could differentiate between straight and gay people, this study came in support of a different theory. This states that people cannot choose to be gay, since there are some hormones which influence this even before they are born.
However, not everybody is happy about the existence of this study. Gay rights militants, as well as the Human Rights Commission (HRC), are worried this might contribute to the increasing stereotypes related to the gay community, and lead to more problems within it.
Ashland Johnson, the director of public education at HRC, condemns Stanford for getting involved in the study, which can be found in the Journal of Personality and Social Psychology.
“Stanford should distance itself from such junk science rather than lending its name and credibility to research that is dangerously flawed.”
Image Source: Wikimedia Commons