 |
Senior Member
|
|
Join Date: Jun 2012
Posts: 19,741
Favourites (more):
BB2024: Ali CBB2024: Louis Walsh
|
|
Senior Member
Join Date: Jun 2012
Posts: 19,741
Favourites (more):
BB2024: Ali CBB2024: Louis Walsh
|
It's been slammed by other researchers
They only used white people
They only used people who identified as heterosexual or homosexual, and male or female on a public dating site
It was only highly accurate when comparing one knowing straight person to one knowing gay person, so it only really shows who of them two were more gay
Also two of the five features used to identify a gay person was their grooming style and their facial expression, I'm guessing that means an unhappy ungroomed guy wil alwaysl be flagged up as quite likely to be straight.
Still the data shows that technology can have a working scale of 'not likely to be gay' to 'very likely to be gay' which is kind of mental. Does make me wonder why they only chose to use white people, and whether it would work on other races.
__________________
|