AI capable of guessing people's sexual orientation by just looking at pictures

Scientists have created a new AI system that uses machine learning to detect the sexual orientation of men and women.

AI or artificial intelligence has grown remarkably in recent years and starting from tech industries to political leaders, everyone is putting a significant amount of importance on the technology. Now a new development in this area has just been revealed. As per this new advancement, now an AI can accurately guess a person's sexual orientation.

A study conducted by researchers from Stanford University has unveiled that a computer algorithm is able to appropriately distinguish between homosexual and straight men-just by scanning photos of their faces. This amazing innovation in the field of machine intelligence was published in the Journal of Personality and Social Psychology. As per a report by The Economist, this new AI was 81% accurate in guessing men's sexual orientation and in case of women, the accuracy level was 74%.

Initially, researchers conducted the test on 35,000 facial images, which people (both men and women) publicly posted on an American dating website.

Researchers Michal Kosinski and Yilun Wang made use of "deep neural networks" and a sophisticated mathematical system in order to analyse this huge dataset, reported Asia One.

Following the experiment, the AI concluded that it was able to distinguish gay men and women because they tended to have "gender-atypical" features, expressions and "grooming styles," which made them appear more feminine.

Also Read: AI invents its own language: Did we humans just create Frankenstein?

The artificial intelligence also took into account several other prominent body features, such as gay men having narrower jaws, longer noses and larger foreheads than straight men.

Humans guessed the sexual orientation of those 35,000 pictures simultaneously along with the AI. The result was quite poor in case of the human judges, as they only managed to accurately detect the sexual orientation of 61% males and 54% females.

Meanwhile, some experts have voiced concern regarding the conclusion that the AI made regarding the distinguishable features of gay men and women. They believe that in the future it might impose stereotypes on the biological origins of sexual orientation and may also breach the ethics of facial-detection technology. Some also claimed that this technology could be used to violate people's privacy and promote anti-LGBT agenda.

Related topics : Artificial intelligence