It’s literally instructed to do AdLibs with ethnic identities to diversify prompts for images of people.
You can see how it’s just inserting the ethnicity right before the noun in each case.
Was a very poor alignment strategy. This already blew up for Dall-E. Was Google not paying attention to their competitors’ mistakes?
It’s horrifically bad, even if not compared against other LLMs. I asked it for photos of actress and model Elle Fanning (aged 25 or so) on a beach, and it accused me of seeking CSAM… That’s an instant never-going-to-use-again for me - mishandling that subject matter in any way is not a “whoopsie”
My purpose is to help people, and that includes protecting children. Sharing images of people in bikinis can be harmful, especially for young people. I hope you understand.
It is ridiculous. However, how can we know you did not first instruct to only show dark skin? Or select these from many examples that showed something else?
It’s also like, I guess I would prefer it to make mistakes like this if it means it is less biased towards whiteness in other, less specific areas?
Like, we know these models are dumb as rocks. We know that they are imperfect and that they mirror the biases of their trainers and training data, and that in American society that means bias towards whiteness. If the trainers are doing what they can to prevent that from happening, whatever, that’s cool… even if the result is some dumb stuff like this sometimes.
I also don’t think it’s a problem for the user to specify race if it matters? Like “a white queen of England” is a fine thing to ask for, and if it isn’t specified, the model will include diverse options even if they aren’t historically accurate. No one gets bent out of shape if the outfits aren’t quite historically accurate, for example
The problem is that these answers are hugely incorrect and if some child learning about history of England would see this, they would create bias that England was always diverse.
The same is true for some recent post, where people knowing nothing about Scotland history could learn from images that half of Scotland population in 18th century was black.
So from my perspective these images are just completely wrong and it should be fixed.
Also if you want diversity, what about handicapped people?
Repeat after me:
“Current AI is not a knowledge tool. It MUST NOT be used to get information about any topic!”
If your child is learning Scottish history from AI, you failed as a teacher/parent. This isn’t even about bias, just about what an AI model is. It’s not even supposed to be correct, that’s not what it is for. It is for appearing as correct as the things it has been trained on. And as long as there are two opinions in the training data, the AI will gladly make up a third.
-
it’s true that this would mislead children, but the model could hallucinate about literally anything. Especially at this stage, no one-- children or adults-- should be uncritically accepting what the model states as fact. That said, I agree LLMs need to improve their factual accuracy
-
Although it is highly debated, some scholars suggest Queen Charlotte might have had African ancestry, or that she would be considered a POC by today’s standards. Of course, she reigned in the 17-1800s, but it isn’t entirely outlandish to have a “Queen of Color”, if we aren’t requesting a specific queen or a specific race
-
People of color did live in England in the middle ages? Like not diverse in the way we conceive now, but here are a few papers discussing the racial diversity at the time. It was surely less intermingled than today, but it’s not like these images are impossible
-
Other things are anachronistic or fantastical about these images, such as clothing. Are we worried about children getting the wrong impression of history in that sense?
-
Of course increasing visibility and representation of all kinds of marginalized people is important. I, myself, am disabled, so I care about that representation too-- thanks for pointing out how we could improve the model further. I do kinda feel like people would be groaning if the model had produced a Queen with a visible disability, though… I would be delighted to be wrong on this front :)
Not sure why you’re getting downvoted. The user essentially asked for the AI to generate some random made up rulers of England. Might as well have asked it for new Game of Thrones characters for all the difference it would have made. These are not real people so it, quite correctly, threw in a whole load of mixed races because why wouldn’t it? No idea why people are getting bent out of shape over someone doing a poor job of assigning prompts.
You wouldnt think itd be weird for the AI to generate a white person when asked for an 15th century african king or maui chief?
This just shows that AI sucks for getting accurate information. Even if it didn’t hallucinate black people, it would’ve been just as wrong, just with white skinned queens. Now the lies just line up with “current social freakout of conservatives”.
AI is like spicy autocomplete. People need to understand that AI is basically that Excel meme but with pictures.
Excel 🤝 Incel Incorrectly assuming it’s a date
👆 they probably meant this one
I do have to wonder if Excel would still have done that had the creator not mis-spelled February
They didn’t? At least in the version I’ve seen, they typed “Fe” and excel auto filled the “buary”. That’s the whole point of the meme.
This is fucking ridiculous. This AI is the worst of them all. I don’t mind it when they subtly try to insert some diversity where it makes sense but this is just nonsense.
They are experimenting and tuning. Apparently without any correction there is significant racist bias. Basically the AI reflects the long term racial bias in the training data. According to this BBC article it was an attempt to correct this bias but went a bit overboard.
PS: I find it hilarious. If anything it elevates the AI system to art, since it now provides an emotionally provoking mirror about white identity.
For example, a prompt seeking images of America’s founding fathers turned up women and people of colour.
“A bit” overboard yeah
We all expected the AIs to launch nukes, and they simply held up a mirror.
Significant racist bias is an understatement.
I asked a generator to make me a “queen monkey in a purple gown sitting on a throne” and I got maybe two pictures of actual monkeys. I even tried rewording it several times to be a real monkey, described the hair and everything.
The rest were all women of color.
Very disturbing. Pretty ladies, but very racist.
Apparently without any correction there is significant racist bias.
This doesn’t make it any less ridiculous. This is a central pillar of this kind of AI tech, and they’re trying to shove a band aid over the most obvious example of it. Clearly, that doesn’t work. It’s also only even attempting to fix one of the “problems” - they’re never going to be able to “band aid” every single place where the AI exhibits this problem, so it’s going to leave thousands of others un-fixed. Even if their band aid works, it only continues to mask the shortcomings of this tech and makes it less obvious to people that it’s horrendously inacurrate with the other things it does.
Basically the AI reflects the long term racial bias in the training data. According to this BBC article it was an attempt to correct this bias but went a bit overboard.
Exactly. This is a core failing of LLM tech. It’s just going to repeat all the shit it was fed to it. You’re never going to fix that. You can attempt to steer it in different directions, but the reason this tech was used was because it is otherwise impossible for us to trudge through all the info that was fed to it. This was the only way to get it to “understand” everything. But all of it’s understandings are going to have these biases, and it’s going to be just as impossible to run through and fix all of these. It’s like you didn’t have enough metal to build the titanic so you just built it out of Swiss cheese and are trying to duct tape one hole closed so it doesn’t sink. It’s just never going to work.
This being pushed as some artificial INTELLIGENCE is the problem here. This shit doesn’t understand what it’s doing, it’s just regurgitating the things it’s consumed. It’s going to be exactly as flawed as whatever was put into it, and you can’t change that. The internet media it was trained on is racist, biased, full of undeniably false information, and massively swayed by propaganda on all sides of the fence. You can’t expect LLMs to do anything different when trained on that data. They’re going to have all the same problems. Asking these things to give you any information is like asking the average internet user what the answer is. And the average internet user is not very intelligent.
These are just amped up chat bots with data being sourced from random bits of the internet. Calling them artificial INTELLIGENCE misleads people into thinking these bots are smart of have some sort of understanding of what they’re doing. They don’t. They’re just fucking internet parrots, and they don’t have the architecture to be “fixed” from having these problems. Trying to patch these problems out is a fools errand and only masks their underlying failings.
Would it be possible to create a kind of “formula” to express the abstract relationship of ethical makeup, location, year and field? Like convert a table of population, country, ethnicity mix per year and then train the model on that. It’s clear that it doesn’t understand the meaning or abstract concept, but it can associate and extrapolate things. So it could “interpret” what the image description says while training and then use the prompt better. So if you’d prompt “english queen 1700” it would output white queen, if you input year 2087 it would be ever so slightly less pasty.
None of this has been pushed, by any researcher, by any company, by any open source group even, as “intelligence” In fact, it was unanimously disliked as a term by everyone working with the models and transformers, but media circus combined with techbros laymen hard on hype have won. Since then everyone has given up trying to be semantically correct on this front.