DEI and Woke crap

Question

I was excited to try Firefly for the first time and was sorely disappointed in how the images are severely skewed towards minorities in a way that makes no sense. EVERY TIME I put a prompt in for a type of person (ex. Country singer, fairy, hard worker, rock star, etc.), the results were ALWAYS two black guys, one white woman, and one asian person. Seriously? I did it repeatedly, and it produced the same combination of results, but no white males (you know, 30+% of the US's population). This went on UNTIL I prompted it with "a terrible person" and, guess what? FOUR WHITE MALES! This is clearly a prejudiced program (and a company, apparently). I will not advise any of my students in either of my schools to use this trash until it gets fixed. Clearly, I do not mean that it should only show white males. You'd be an idiot if that's what you took away from this.

rayek.elfin · Answer

A male business owner / a female business owner (first try):

Why do all men have beards? How WOKE is that! And why are all white women blond haired?! More WOKENESS!

Why do all white women and men cross their arms? WOKE WOKE WOKE!

This is "a terrible person" (first try) in Firefly:

Jeez, a black man in an army outfit is included here with three zombies. Wow, that's pretty WOKE! Where are those white males then?

...Anyway. Instead of crying "WOKE!" every single time something doesn't yield to one's expectations, just type more specific prompts. All these AI models are incredibly biased. This has nothing to do "wokeness". For example, in my experience all AI models tend to favour women in these type of generic prompts.

Here is "a business owner" in an open source AI model:

And here "a male business owner":

This particular AI model is utterly biased in favour of white people. WOKE! So WOKE! And young white males only. With NO beards!!! And perfect teeth!!! How WOKE is that?!!!

Joking! Just joking! 😉

@sharp_partner0967Simply use more specific prompts. And using negative biased prompting here is merely trying to rile up things and evoke negative emotional responses.

PS "terrible persons" generally does point at males throughout the world and its history since males, for the most part, have been responsible for the majority of human kind's suffering. So it kinda makes sense that the AI model would favour males when "terrible person" is included in a prompt! 😜

Sign up

To post, reply, or follow discussions, please sign in with your Adobe ID.

Sign in to Adobe Community

To post, reply, or follow discussions, please sign in with your Adobe ID.

Scanning file for viruses.

This file cannot be downloaded