Hey, if you want to argue with the parameters about what stereotypes come out most strongly for that prompt, be my guest. But no need to shoot the messenger. LLMs are basically stereotype machines, it just turns out if that you make a trillion stereotypes, you get a lot right.