It's almost always telling isn't it?

Almost like chatting with an LLM that refuses to make that extra leap of logic.

"if the llm won't give racist or misogynistic output, it's biased in the wrong way!"