AI can detect sexuality of people based on photo, study reveals

Critics slam the researchers of the study, calling it 'junk science'.

A group of researchers has been met by criticisms after publishing a study on an artificial intelligence technology used in a dating website to detect whether an individual is gay or not based solely on photos.

A study from Stanford University has shown that a computer algorithm is capable of making a distinction between straight and gay men 81 per cent of the time and 74 per cent for women. The study, published in the Journal of Personality and Social Psychology, used 35,000 close-up photos of men and women publicly posted on a dating website.

Also read: French software firm selling 'gay-detecting' software faces backlash

Researchers Michal Kosinski and Yilun Wang used the so-called "deep neural networks", a complex mathematical system that knows how to examine images, to extract features from the photos.

According to their findings, trends among 'detected' gay men appear more feminine in their expressions and grooming styles. Additionally, they have narrower jaws, longer nose and larger foreheads than straight men.

The study also supported the theory that sexual orientation roots back to particular hormones before birth, suggesting that being gay is not a choice but innate.

While many find the study results interesting, critics are not happy about the whole process. Popular LGBTIQ groups in the US, the Human Rights Commission (HRC) and Gay and Lesbian Alliance Against Defamation (GLAAD), have condemned the study, worrying of its flaws and what it can contribute to the growing issues affecting the community.

"Stanford should distance itself from such junk science rather than lending its name and credibility to research that is dangerously flawed and leaves the world – and this case, millions of people's lives – worse and less safe than before," says HRC's director of public education and research Ashland Johnson in a statement.