A group of Stanford University researchers have come under fire after creating an AI bot they claim can distinguish between different sexualities.
An experiment they conducted, based on the facial features of people signed up for a US dating site, was published in the Journal of Personality and Social Psychology.
The study claimed its bot was able to correctly identify the sexuality of 81 percent of males on the site, and 71 percent of women.
"Gay faces tended to be gender atypical," the research stated. "Gay men had narrower jaws and longer noses, while lesbians had larger jaws."
However in a follow-up test, the software did not perform as well - and was bagged by The Economist, which said the research had a number of "limitations".
And it was in that publication that the research was spotted by two LGBT rights groups, who issued a joint press release describing the study's findings as "reckless" and "a weapon to harm".
The Human Rights Campaign also criticised the research, and said the university should move to disown the study.
"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," director of research Ashland Johnson said.
Professor Michal Kosinski and Yilun Wang, who headed up the study, have now hit back, saying only "scientific findings" are an appropriate rebuttal to their findings - not "well-meaning lawyers and communication officers lacking scientific training".
Newshub.
