Thursday, April 2

Privacy Concerns: Artificial Intelligence can now Detect Your Sexuality

Google+ Pinterest LinkedIn Tumblr +

A new study has revealed that artificial intelligence can now detect our sexuality. Is this a threat to our privacy? Is it time to disconnect?

The field of artificial intelligence (AI) is evolving rapidly. With a keen eye for detail and sharp perceptive qualities, advances in artificial intelligence have led to the detection of patterns and even the development of its own “gaydar” system. AI has now been afforded with the power and ability to infer sexual orientation merely by analysing people’s faces. The researchers, Michal Kosinski and Yilun Wang, Stanford University, argue that the software does this by recognising subtle and delicate differences in facial structure.

The study that the software was relied upon for, processed 130,741 images of 36,630 men and 170,360 images of 38,593 women downloaded from a popular American dating website (which has not been revealed for copyright purposes) which makes the profiles public. The software identifies the ‘faceprint’ of each person and uses a simplified predictive model, known as ‘logistic regression’, to discover correlations between the faceprints and the individual’s sexuality.

One photo was chosen at random each time and the model could correctly identify if the person in the photo was gay or straight, with 81% accuracy. When shown five random photos of each man, it identified the individual’s sexual identity correctly 91% of the time. In women, the model told straight and gay apart with less precision. However, after one photo, it attributed sexuality with 71% accuracy, and after five photos with 83% accuracy.

Percentages may reveal one thing, but the interpretation of such statistics tend to reveal a lot more as the research found that gay men and women tended to bear “gender-atypical” features, expressions and grooming styles, meaning that gay men physically appeared more feminine and gay women physically appeared more masculine. These physical attributes included facial structure, jaw lines and cheekbones, forehead size, and nose sharpness, among others.

Artificial Intelligence, sexuality

From shape of the eyes to the presence of a smile, software detects a great deal more than just colours


This study begs us to ask the question: is sexual orientation biologically determined? The findings suggested that there is strong support for the theory that sexual orientation is determined by an exposure to various hormones in the developing stages preceding birth, indicating that one is born gay and does not choose to be queer or heterosexual. The adage that is used by some, that queer individuals make a choice to be queer, is simply false and many studies have been performed to reveal whether a correlation between biology and sexuality exists.

Some argue that it is social constructionism that shapes one’s gender or sexual identity. However, nature and nurture work alongside one another and to reject one is to reject the other. Behavioural changes, for instance, also have a biological explanation, and our response to socialisation and cultural norms can be traced back to our genetic markers and biological predispositions. Sexual orientation is, by default, a pattern of desire and not just a pattern of acts or behaviours. And while gender fluidity (predominantly among women) can be used as an argument against the biological foundations of sexual identities, we are each built to have different desires and attractions. Studies like these could potentially put an end to this debate.

“Nature and nurture work alongside one another and to reject one is to reject the other”

While the software developed by Kosinski and Wang has accurately identified between this specific sample of gay and straight individuals, it still has not been fine-tuned to recognise other sexual orientations. People of colour were not included in the study and transgender and bisexual individuals were not considered in the sample group either. This not only raises concerns surrounding the limitations of the validity and efficacy of such software but it also raises questions revolving around the alarming implications of artificial intelligence. More importantly, it causes us to interrogate the ethics of facial-detection technology and whether this field of work and type of software violate our privacy.

Facial recognition and the power of machine vision is now in full force and yet, we have been subject to it already. Similar controversial studies have already been executed. Dr Kosinski has repeatedly played with fire and was indeed the scientist behind psychometric profiling that used Facebook data to utilise information displayed on a person’s profile to archetype their personality. Even governments use public data to track a citizen’s interests. With social media overflowing with profiles, facial images, and personal information, could the access to public data be used to detect people’s private concerns (e.g. sexuality) without their permission?

What does this mean for our privacy? As it is a major feat and advancement for the worlds of science and technology, this could have a serious impact on societies and communities across the globe. This sort of technology comes with certain dangers and risks and artificial intelligence could be used to humiliate, out, other and persecute members of the LGBTQI community. In countries where homosexuality or anything straying from heteronormative values and traditions is considered illegal or immoral, this software could pose a serious threat to those individuals’ safety as they live under the radar. Hate crimes against members of the LGBTQI community could skyrocket even more if this kind of technology is released for public use in the future.

Artificial Intelligence

AI doesn’t and won’t stop there. It could track your whereabouts, address, phone numbers and maybe even convict one of a crime solely based on a conversation on Facebook. Privacy tools must be put into full swing effective immediately before machine learning and AI become misused and misinformed.

Our computers and phones are becoming our new best friends; tracking our every like, analysing our every comment. They’ve come to know us better than our own family and sometimes are even a step ahead of us by discretely using ‘cookies’ to reel us to the cookie jar of information gluttony.

We’ve come so far and it is more possible to imagine the end of the world than the end of technology itself. Privacy is slipping from our fingertips as we enthusiastically type and click on the daily. But would you stop using technology if you knew your privacy was in jeopardy? Guess what, it already is.



About Author


Dominic is a Greek/American writer & editor; English and Theatre Studies Graduate

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.