[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/pol/ - Politically Incorrect


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


Picrel


This sounds like schizo shit but hear me out

So I assume that unless you literally live under a rock you are familiar with the fact that AI image and video generators have rapidly developed in terms of succeeding in appearing hyperrealistic in quality (see;sora 2 by OpenAI, veo3 as video examples. Grok also produces 6 second videos on Grok Imagine)


This of course leads to a lot of degenerate racemixing coomers using the more unfiltered/less heavily censored sites for producing pornography and shit but thats besides the point here

But one thing thats noteworthy is that, when you want to depict people in AI images or videos...unless you very explicitly and directly mention 'white' man/woman in your prompt (especially with the latter)...

It will default to producing very vaguely Latina or Asian-adjacent mystery meat images. Like if you state 'fair-skinned attractive young woman' [does/says insert generic innocuous thing], it will default to depicting very obviously Asiatic features lmao

Even Grok, which is supposedly Elon Musk's 'based redpilled uncensored' AI app, is not at all immune to this and for some reason is actually more blatantly prone to this than most.

While I obviously am not some porn-addicted degenerate, I decided to see out of curiosity what happens when you ask AI to produce a very 'generic' or 'lazy' prompt that some horny coomer would write - 'sexy cute woman' 'attractive/elegant/hot woman'.

My mental image would be..tall, slim, pale-skinned, straight-haired European woman with long, slender legs and sharp, angular facial features. Basically what a 2004 European model looks like.

...this is what you get in picrel. Very obviously ethnically ambiguous and not standardized European features. It defaults to anti-white results.

Whats your take anons? How do you explain this?

(Ignore my convoluted writing style in picrel thats just how i always communicate with ai)
>>
>>539062774

It literally took like a 80-120 term hyperspecific prompt to grok to depict my mental image of an attractive woman.

Why dont women irl look like this anymore?

I had to edit prompts over and over again with like 100-120 terms for hyperspecific details to get an actually tolerable vibe that isnt racemixing or degeneracy adjacent and looks like an actual attractive white female


Why dont women look like this?
>>
But I thought /pol/ said AI was going to be based and tell the truth about race and gender.

What happened?
>>
>>539063051

Only uncensored.ai actually gets that shit but sadly it cant create images
>>
>>539062874
Listen Space Force, this is alien speaking.

You are not doing good enough.

You already posted that image and whined about "why are there no women like this anymore"

You are so lazy it beggars belief
>>
>>539062774
Gemini on launch
/thread
>>
>>539062774
it was also observed that AI overlays nigger faces on to everything.
for example, someone prompted AI to recreate pictures of white people, and repeated it a 100times. with every iteration, the white persons turned more into niggers.with a single iteration alone you could notice that nigger features were added to their faces.
>>
>>539062874
Because everyone is brown
>>
>>539062774
It responds to reactions and countinued use, considering there are far more asians and blacks in the world it makes sense that these grooners would skew the data toward there race. We need racially segregated ai because it is mixing racial psyches
>>
>>539062774
China took the lead with image generation models so the majority of women in the data sets that were used as the basis for most models was biased toward China.
>>539066893
My understanding of how image generation works is that models do not update in real time which means that nothing users do matters and it's all about training data. As a disclaimer, I only fuck around with local models and generation because I don't trust the kikes that do it server side but I assume it works the same way. /g/ would probably know better.
>>
Yes, this is widely known and discussed.
>>
File: H0171-L253321941_original.jpg (296 KB, 2146x1500)
296 KB JPG
>>539062774
Yes

>>539062874
Because jews hate beauty
>>
>>539062774
>Elon Musk's 'based redpilled uncensored' AI app
Yeah ...
Out of the 50+ times that shit site has banned me, one was for using Grok`s own data about holohoax
>>
>>539062774
I just convinced an AI (not gonna say which) that 0=1 and then asked it to code the formula (it was more advanced math) and it went into a recursive loop and I left it running on my phone for like an hour and a half while I ate pizza and thought about how I am 100 percent certain that 1=0.
>>
>>539062774
I’m white but mostly jerk off to Asians and Latinas. Maybe grok just better understands what coomers want.
>>
File: IMG_7467.jpg (240 KB, 1169x1928)
240 KB JPG
>>539067123
>he doesn’t want his data juice in the data punchbowl
That’s no fun
>>
>>539062774
AI defaults to Asians when it can’t generate “young” looking women due to restrictions because Asian women are generally smaller and appear younger than their age relative to other races.
>>
>>539062774
because most people would find this more attractive over >>539062874
>>
>>539062774
Kek
>>
>>539062874
>>539062774
MAKE THEM DRINK PISS



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.