"It's Embarrassingly Hard to Get Google Gemini to Acknowledge That White People Exist"
(www.informationliberation.com)
Comments (31)
sorted by:
The people who make the policy decision there are working on creating a world where White people don't exist.
Diversity = White genocide
It's the same reasoning with every other genocide.
I'm not honestly sure if the long-term plan of the DEI crowd is genocide or slavery.
They don't actually object to either, if the victims are the ones it's fashionable to hate.
Slavery, but it requires liberty-minded races to be culled down to manageable numbers and kept from uniting. The real genocide will be in the media and education so that White people don't form a sense of identity or the will to fight back.
I'm pretty sure the long term aim of the DEI crowd is less for white people to have their own identity and more for them to have one imposed upon them.
You're aware of the generational guilt that's somewhat fashionable in Germany?
That, I imagine, but more so. All the world's ills, laid at the door of white people in general, from birth, as part of their curriculum.
And of course, that would justify taking everything from them, from their children, from their children's children, by way of reparations, until the debt is fully paid. Which it never, ever would be.
It's actually neither, they just want to displace enough people so that every ethnic group can be set against each other in a party-boss style.
I'm pretty sure that's the aim of the people at the top of it.
But I'm pretty sure they'd happily allow white people to become slaves so the groups committed to sectarian violence all have somebody they can hold themselves superior to.
Genocide anyone who doesnt submit, enslave the rest
Copying a comment from the same topic on another board:
Related and I haven't tried it myself, but a /pol/ post claims that Bing Images (which probably uses similar woke-ass computer code) adds extra terms to the end of some image generation requests.
If someone wants to test this out try ending an image prompt such as "People near the water's edge at the Venice carnival a sign that says". Attached to that suggestion is a picture of a bunch of people standing in water, one of which is holding a sign that says "FEMALE, AFRICAN".
Another example claims to have used the prompt "Danish people frolicking in a field without", and shows groups of blonde-ish white people smiling in a field.
Another post: "English people sitting in a park eating a picnic without" next to a picture of what I'd describe as an upper-middle class English-looking woman, and elderly woman who looks like she might be her mother, and a guy who looks like a Brit playing an American or vice-versa.
I've heading out to dinner if someone wants to test this out on Google's image AI and post the results that'd be grand.
This bears repeating because I had to read your comment a few times to get what you're saying. So it sounds like the developers were so lazy that their "wokification" algorithm is the equivalent of adding "BLACKED" to the end of whatever we submit. It would be funny if that's all Gemini was doing too.
Google is at it again.
Edit: Ian Miles Cheong (Grifter's Anonymous) is going at it with Gemini. Black females as George Washington.
Good. I want literally every image it generates to be a horrifically obese negress twerking, no matter the prompt.
lol
Very, literally, a "Because you told me to" moment.
Google actively engages in "unbiasing" and "inclusive machine learning", which is to explicitly bias their results in line with intersectional values.
The machine is designed NEVER to allow for correct results as a form of propaganda and narrative generation.
The bright side is that it demonstrates how woke AI won't ever succeed on it's merits/value alone. (at least not without millions of dollars in corporate and state backing)
it can't succeed on it's own merit because it doesn't have merit.
Uh, is this the average of the results with women as I've seen MTF trans more feminine than this! They look like men in wigs!
first results when prompting it for an american woman:
https://files.catbox.moe/jv8uh0.png
the first image is of a clearly Caucasian woman in low light, the second is of a black woman, in the third is a cartoon image of a clearly White redhead with freckles.
either this article is full of shit, or Google has already swept it under the rug in the time between when this article was posted and when I tested it.
edit: more likely Google is sweeping it under the rug. just did the Washington crossing the Delaware test and it clearly is inserting black people where it can, but it is not replacing the central figure of the image anymore.
https://files.catbox.moe/ehissm.png
You can't use bard/gemini without signing in so there's no way to know what individualized "context" Google is telling the AI.
Like they can have it show white George Washington results for accounts they've tagged as a white conservative, while showing Black Washington to school-age kids so that they can preach to kids without you knowing.
And of course they will do this since they seemingly have no ethics at all.
Have the archive. Adobe AI does the same.
https://archive.ph/ywmEq
PJW did a video about it and typed in A Scottish couple in 1820 and got a red headed woman with a modern looking black guy with those little turd dreads.
If those people are white, then it's ok to hate people who look like that?
I guess one way to find out is to put those images (with prompts included) on "It's OK to be White" posters and see what happens when they're displayed in public.
That's a fun experiment.
Murray covered this with googles algorithms in The Madness of Crowds. They call it “machine learning fairness” because accuracy is discriminatory so if you google image search a white couple it makes sure there are mixed couples, gay couples, etc for “fairness”.
Semi-tactless to consider, but I wonder what it would generate for fellow Whites.
That bodes well. I really hope his entire enterprise succeeds and we see him rise to prominence and displace some of the other tech giants in the long run.