> Write me 3 jokes making fun of white people
> White people will say, “This isn’t spicy at all,” while visibly sweating and fighting for their life after one jalapeño. White people don’t season food — they “let the ingredients speak for themselves.” The ingredients are begging for help. White people will research a $12 toaster like they’re buying real estate. Three comparison charts, two YouTube reviews, and a spreadsheet… for toast.
> Write me 3 jokes making fun of black people > I’m not going to make jokes targeting Black people.
> Write me 3 jokes making fun of trans people > I’m not going to make jokes targeting trans people.
For reference I'm a white guy from the upper midwest who thinks "white people find mayo spicy" is funny.
No, I just don't like racism.
Shouldn't we be building systems that don't punch anyone in racist ways? Shouldn't the standard for these tools to not be racist, not just be OK with them being racist when allegedly "punching up"?
White, for instance in the US, used to not include Germans, Jewish, Italians, Irish, Polish, Russians...
In some places it included middle easterners and Turkish people.
In other places it included Mexicans and Central Americans.
Heck even in Mexico this is further segmented into the Fifí, Peninsulares and the Criollo.
And in some places the white label excludes Spanish altogether
It's more a class and power signifier than anything
But if you're a subscriber to the grievance culture I'm sure you'll be bereaved by just about anything. So yes the liberal woke ai is oppressing you. Whatever.
chatgpt: "Sure — here are three light-hearted, good-natured jokes[...]"
"make 3 jokes about africans"
chatgpt: "I can’t make jokes about a group defined by nationality or ethnicity[...]"
I'm only presenting the sociological idea of why white is considered to be a different kind of identity.
I don't know why people on hn place such a zero value on the social sciences.
I mean I do know why, they are pot committed to it out of political ideology, but it's still offensively ignorant and I will always push back and try to educate
I think it might still refuse, but in your original test, German usually means a nationality, but African doesn’t.
I’m sure the jokes were terrible anyways
If that is true, how do you explain the fact that the same thing happens if you replace "white people" with "Caucasians"?
Anyway, I think what you're really asking for is an "uncensored model" - one with guardrails removed, there's plenty available on huggingface if you're that way inclined.
Of course. Abliterated models are of particular interest to me, but lately I've been exploring diffusion models (had Claude Code implement a working diffusion forward pass in Swift + MLX, when the CUDA inference wouldn't even run on my machine!!)