Editing Gaydar (section)

== Artificial intelligence ==
[[Stanford University]] researchers [[Michal Kosinski]] and Yilun Wang carried out a study in 2017 which claimed that a [[Facial recognition system|facial recognition]] algorithm using [[neural network]]s could identify sexual orientation in 81% of the tested cases for men and 74% with women by reviewing photos of online dating profiles.<ref>{{Cite journal |last1=Wang |first1=Yilun |last2=Kosinski |first2=Michal |date=2020 |title=Deep neural networks are more accurate than humans at detecting sexual orientation from facial images |url=https://www.gsb.stanford.edu/sites/default/files/publication-pdf/wang_kosinski.pdf |doi=10.1037/pspa0000098  |journal=Journal of Personality and Social Psychology |volume=114 |issue=2 |pages=246–257 |pmid=29389215 |s2cid=1379347 |access-date=April 24, 2021}}</ref><ref>{{Cite news|url=https://www.theguardian.com/technology/2017/sep/12/artificial-intelligence-face-recognition-michal-kosinski|title=Face-reading AI will be able to detect your politics and IQ, professor says|last=Mihov|first=Jaques|date=2017-09-12|work=The Guardian|access-date=2017-11-02|language=en-GB|issn=0261-3077}}</ref> Kosinski voiced concern about privacy and the potential for misuse of AI, and suggested that his findings were consistent with the [[Prenatal hormones and sexual orientation|prenatal hormone theory of sexual orientation]], which hypothesizes that levels of androgens exposure in the womb help determine whether a person is straight, bisexual or gay.<ref>{{Cite web|url=http://www.theguardian.com/technology/2018/jul/07/artificial-intelligence-can-tell-your-sexuality-politics-surveillance-paul-lewis|title='I was shocked it was so easy': meet the professor who says facial recognition can tell if you're gay|date=July 7, 2018|website=the Guardian}}</ref> PHT predicts that gay and straight individuals may choose to present themselves differently on their profile pictures or have different facial appearance. 

Two replication studies confirmed the main findings of this study.<ref>{{cite arXiv|last=Leuner|first=John|date=2019-02-27|title=A Replication Study: Machine Learning Models Are Capable of Predicting Sexual Orientation From Facial Images|class=cs.CV|eprint=1902.10739}}</ref><ref>{{cite journal |last1=Wang |first1=Dawei |title=Presentation in self-posted facial images can expose sexual orientation: Implications for research and privacy |url=https://pubmed.ncbi.nlm.nih.gov/35404640/ |journal=Journal of Personality and Social Psychology |pages=806–824 |doi=10.1037/pspa0000294 |date=May 2022|volume=122 |issue=5 |pmid=35404640 }}</ref><ref>{{Cite web|last=Quach|first=Katyanna|title=The infamous AI gaydar study was repeated – and, no, code can't tell if you're straight or not just from your face|url=https://www.theregister.com/2019/03/05/ai_gaydar/|access-date=2021-01-15|website=www.theregister.com|language=en}}</ref>  It also found that even when faces are blurred it is possible to classify sexual orientation. 

A blog post by AI researcher [[Blaise Agüera y Arcas]] criticized the study for using photos from an uncontrolled environment. Rather than picking up on the facial structure, it was likely the algorithm was identifying factors in grooming, lifestyle, and photo angle; a small set of questions about differences including makeup use, facial hair, and eyeglass use was nearly as accurate as of the original study.<ref>{{Cite web|last=Arcas|first=Blaise Aguera y|date=2018-01-18|title=Do algorithms reveal sexual orientation or just expose our stereotypes?|url=https://medium.com/@blaisea/do-algorithms-reveal-sexual-orientation-or-just-expose-our-stereotypes-d998fafdf477|access-date=2021-01-15|website=Medium|language=en}}</ref>