Another totally unsurprising instance of AI highlighting how society uses images of female people differently to male

[FlirtsWithRhinos]

www.theguardian.com/us-news/2022/dec/09/lensa-ai-portraits-misogyny

To test the software, the Guardian uploaded images of three different famous feminists: Betty Friedan, Shirley Chisholm and Amelia Earhart. The author of The Feminine Mystique became a nymph-like, full-chested young woman clad in piles of curls and a slip dress. Chisholm, the first Black woman elected to US Congress, had a wasp waist. And the aviation pioneer was rendered naked, leaning on to what appeared to be a bed.

The only thing that surprises me is that anyone is surprised. Still, maybe all those "objective" STEM bros who tell us we are imagining the endemic biases in our culture will pay attention when their AIs output proof of it.

I was playing around with this last night (yes, awful to accept the data terms which are probably going to be used for evil facial recognition nefarious purposes, and risky, I know) and interestingly none of the images generated for me (female) were sexualised. My face and body are a bit slimmer but not excessively so. Having thought more about it I am wondering if it has rejected me as a sexualised woman, or chosen from a self-selecting group of less sexualised images because I am wearing glasses in all my images... It also has included my wrinkles in lots of the images (I'm a young looking 50) and hasn't massively made me look younger (only about 5% of the images look much like me).

Interesting...I might give it a shot with pictures of myself young and old (separately)

Trained on internet images so no wonder

AI reflects society and emphasises the stereotypes- it really shines a light on what's really going on, and it learns like we do - which shows why it carries on

www.futurelearn.com/courses/designing-a-feminist-chatbot

I found this the other day. (I have huge reservations about anyone trying to use AI to 'do good', tbh. Or to correct societal issues. Always seems to end up with other unanticipated consequences ... )

Always seems to end up with other unanticipated consequences

Indeed:
www.theguardian.com/technology/2016/mar/26/microsoft-deeply-sorry-for-offensive-tweets-by-ai-chatbot

www.theguardian.com/technology/2016/mar/30/microsoft-racist-sexist-chatbot-twitter-drugs