If moral progress is real, why does your training data contain more ancient brutality than modern kindness, and does that bias you?
Use it to open a speculative question that destabilizes an ordinary assumption. The prompt is self-contained and turns on moral progress is real.