deleted by creator

I’m not exposed to a huge amount of media coming out of Asia, outside of a handful of Korean shows that Netflix has picked up and anime. But like, if anime is any indicator, I’m not really surprised that the training data for Asian women is leaning more toward overt sexualization. Even setting aside the whole misogynistic ‘fan service’ thing, I don’t feel like I see as much representation of women who defy traditional gender roles as the last twenty or so years of Western media.

It certainly could be that anime is actually a huge outlier here, but if the training data is primarily from the English speaking web, it might be overrepresented anyway. But like, when it comes to weird AI image behaviors, it pays to think about the probable training data.

Like, stable diffusion seems to do a better job of rendering jewelry if you tell it to surround it with berries. Given the output, this seems to be due to Christmas themed jewelry ads. They also tend to add a lot of bokeh for the same reason.

If I had to guess, they probably did a shit job labeling training data or used pre labeled images, now where in the world could they have found huge amounts of pictures of women on the internet with the specific label of “Asian”?

Almost like, most of what determines the quality of the output is not “prompt engineering” but actually the back end work of labeling the training data properly, and you’re not actually saving much labor over more traditional methods, just making the labor more anonymous, easier to hide, and thus easier to exploit and devalue.

Almost like this shit is a massive farce just like the “meta verse” and crypto that will fail to be market viable and waist a shit ton of money that could have been spent on actually useful things.

Stable Diffusion is little more than content laundering. It cannot create anything more than what you put in.

Looking at some of the replied that tried to dismiss the issue and the general lack of concern from moderators against aggressive replies from AI apologists (in this thread but also other AI related threads) are disheartening.

Wrong question. The right question would be:

Why is AI as used in Lensa’s Magic Avatars App Pornifying Asian Women?

Ask Lensa to remove the “ugly” and similar negative prompts from their avatar generating App, and let’s see what comes out.

https://stable-diffusion-art.com/how-to-use-negative-prompts/#Universal_negative_prompt

For reference, check out how that same negative prompt turns a chubby-ish poorly shaved average guy, into a male pornstar, or a valet into a rich daddy’s boy.

Because white dudes fetishizing asian women wrote the llms and pointed at the training data