Lesbian dating app to use facial recognition to exclude males

[zibzibara]

https://www.dailymail.co.uk/news/article-13484121/Lesbian-dating-app-use-facial-recognition-exclude-trans-women-matching-biological-females.html

This is from Jenny Watson who's known for fearlessly kicking out the males invading her lesbian speed dating events, now she's doing the same with a new dating app for women. Hopefully it will work as advertised but even if the usual crop of males manage to impose themselves through whatever deceitful trickery, it just further proves her point about how lesbians are being bullied out of their own spaces. Wishing her the best of luck with this!

I think they're really good! Scarily so...

With the B2C ones, or commercially available B2B ones that you may use in your work, I appreciate they may not be as accurate, but remember Moore's Law - they're getting better as we speak. I think your concerns about race are a bit overblown tbh, and kind of...phrenology coded? That makes me a bit uncomfortable. But I appreciate it's coming from a good place.

With something like the dating app, I think a user could use the software to support any suspicions they may have. So if you're a lesbian and only want women, and you get matched with a person who you think might actually be a man pretending to be a woman, then you can use the software to increase or decrease the likelihood of your suspicions being true.

I mean, humans already have a pretty decent facial recognition software built into our brains - we've been visually distinguishing males from females for longer than we've been walking fully upright! So I guess it would be able establishing a probably that the woman you're talking to is actually a fake, and then you can swipe left and move on to the next profile.

With the B2C ones, or commercially available B2B ones that you may use in your work, I appreciate they may not be as accurate, but remember Moore's Law - they're getting better as we speak. I think your concerns about race are a bit overblown tbh, and kind of...phrenology coded? That makes me a bit uncomfortable. But I appreciate it's coming from a good place

I can accept that they’ll get better really fast. I don’t totally understand the phrenology coded criticism sorry!

I vaguely remember hearing this was an issue in the earliest version of Sall Grover's app but was subsequently ironed out. Ultimately comes down to the AI being fed a disproportionate number of white faces I assume, just as AI in general suffers from being exposed to too little/biased source material about women and thus in turn resulting in algorithmic bias.

So phrenology was a 'science' that racist white men used in centuries past. They believed that Black folks had fundamentally different skull shapes and that Black Women were more masculine and not as feminine as white women.

This vile ideology was used to justify the rape and abuse of Black Women during slavery and also to justify racial laws and discrimination. Basically, it was this idea that Black folks were less 'evolved' than white people and that their skulls could prove it.

It's unbelievable bollocks of course that has been disproved many many times. But as a Person of Colour, when I see comments alluding to the idea that AI might have difficulty 'recognising' Black people's faces, for me there's an underlying idea beneath it that Black people's bone structures mark them out as fundamentally different and less 'evolved' than Causacsians.

It's disgusting.

now I appreciate that you might not be aware of all that history, but those are the racist assumptions that lie beneath chat about Black people's faces, skulls, facial features etc. You probably didn't even realise that there was a racist idea beneath it and I am NOT calling you racist lol! Just thought it was something you might find useful to be aware of in future, especially if you work in media, which definitely needs more Black representation, especially in adverts.

Can I ask if you work in a big advertising agency?

This gave Dylan milvany (two separate pictures used) more % of accuracy than of me. An actual born woman.

It's not phrenology at all! It's about western standards of femininity being the cultural standard globally and the negative impact that has on some women (look at comments many people make about the Williams sisters for instance).
Archaeologists will never tell you that a skeleton is 100% anything. Many older statements were based on height assumptions for instance.
AI facial recognition is like LLM. Only has good as the data it's based on. And that comes with a huge amount of human based assumptions.
Fortunately, as far as I'm concerned, I couldn't care less in this specific instance as I disagree with most people on this board. However, this will not do what people think. And will exclude plenty of women for no reason other than they don't meet Western standards. The so-called skeleton based issues are also used by 'trans-hunters' who end up attempting to 'out' half of Hollywood as being trans women! It's genuinely ridiculous.

Can I ask if you work in a big advertising agency?

No I work in news media, not advertising.

Bumping this.

A simple, 100% accurate test should be to ask the person what sex they are.

We don't even consider that because we know some men will lie. They believe their needs and wishes are more important than those of the women they want to date.

Furthermore, cross sex identity activism has bastardised our language so much there isn't even a word that everyone agrees includes all female people and excludes all male people.

So people end up getting angry about the fairness of tools we only need in the first place because men cannot be trusted, when it's the latter that we should be really angry about.

When I mentioned it, it was the fact that the AI is trained on historical beauty standards.
It will also reject more athletic women.
It's trained on what exists and because the media tell us feminine = delicate, dainty white women you see that DM gets a higher 'woman' rating than Whoppi Goldberg for instance or the actress who plays Brienne of Tarth. It's an issue because the data fed into it is racist (and sexist). And you're right that's not on.
For a whole host of reasons we need better representation of all different types of femininity in the media - I like to think that's something everyone on here could agree with?

Isn't it just awful that such technology is evennecessary and all because some men won't take no for an answer.

Where did you get these images from. And are you confident that they haven't already been digitally manipulated?

The software we use at work is trained on osteobiological data. I don't know what the lesbian dating app software was trained on.

Is it not in everyone's interests to feed in a hell of a lot more faces to counteract the problem? I don't understand why this is so hard.

Or to have databases and programs for different ethnicities, which would be more reliable?

Surely the Chinese have a phenomenal one based on their population!

In terms of our own skills, we don't work as well on photos, we work on 3D information- including movement.

I don't think facial recognition is something that should be used at all!

Lucky you to have both Whoopi Goldberg and Dylan Mulvaney beside you to use the camera on the app to verify.

Even Tinder used verification via camera only, matching their silly poses and making sure your head fit into their oval. Pretty sure this app won't accept photos to verify you are a woman.

Hopefully they also won't allow the original uploaded pic to be deleted once verification is complete so other pics can be uploaded instead (Tinder did) but not sure how much of a problem actual women catfishing other women is 🤷

Well exactly. All these complaints of feminity policing and genital inspections fails to put the responsibility at the right door. That of gender ideologists who have told men, not only that they have a right to females places, but that they are righteous to claim these spaces as their own.

Until gender ideology came along, men did keep out of our spaces.

Archaeologists will never tell you that a skeleton is 100% anything. Many older statements were based on height assumptions for instance

I’ve spoken to an archaeologist about this as I’ve seen this claim here before. He’s worked for over three decades and skeletons are rather his thing. He says that yes, you can tell whether an adult skeleton is male or female. He says sometimes you get more masculine looking female skeletons or feminine male ones but you can always tell the sex. He says for adolescents it’s not always clear and you can’t tell the sex with children, but for adults he said yes, you can tell the sex from the skeleton.

DM has manipulated his face through surgery. Literally had bits carved off his skull. That’s got to confuse the algorithm.

Most men trying to access the site won’t have had that.

From what Sall Grover has said the app she used works pretty well (though facial surgery may make these apps less reliable) She is in legal trouble just for excluding males who identify as women from her lesbian dating site because Australia thinks that's discrimnination. She may have to go all the way up the Australian legal system and no guarantees she'll win.

Well it’s a lesbian app so if it polices femininity it’s going to spectacularly fail isn’t it?

I think Sall Grover said she set the accuracy at 95% which means it lets in all women and a few males. The males were then removed manually by a human. Sadly there will always be a few abusive men who slip through the net though.

Got to say I'm feeling pretty uneasy about the posters @GrammarTeacher and @DuchessNope who seem to be implying or stating outright that Black Women have more masculine features because they are Black??!?

I haven’t said that anywhere. I’ve said that facial recognition apps seem to think they do.

To reassure @CantDealwithChristmas that absolutely no one thinks or is suggesting that black women are ‘less beautiful’ or that comments alluding to the idea that AI might have difficulty 'recognising' Black people's faces, for me there's an underlying idea beneath it that Black people's bone structures mark them out as fundamentally different and less 'evolved' than Caucasians.

To be clear, the concerns about AI having difficulty recognising black faces such as mentioned by @GrammarTeacher arises from the many research studies on FRT. For example, MIT computer scientist Joy Buolamwini has written extensively on this issue.

My guess is that anybody questioning AI’s sensitivity to misidentifying black faces comes this understanding. Kate Crawford wrote about this in her book Atlas of AI and how it comes from a lack of representation in AI’s training data. This then seems to have been corroborated by stories irl of black men being wrongly arrested on the basis of (poorly designed) FRT.

However, going by what you say here, it is at least reassuring that the UK border controls use more reliable technology.

https://www.media.mit.edu/articles/facial-recognition-software-is-biased-towards-white-men-researcher-finds/

https://mitsloan.mit.edu/ideas-made-to-matter/unmasking-bias-facial-recognition-algorithms

https://www.nytimes.com/2020/12/29/technology/facial-recognition-misidentify-jail.html

I’ve also read that some facial recognition software has a racial bias. Apparently it recognises white men with the highest accuracy and black women with the lowest.

No facial recognition needed if men just stay off a site set up for women who are attracted to other women.