[frumplestlatz]

I wonder how much the harm-based ethical model routinely applied in the name of AI "alignment" limits models' capacity for humor.

It's hard to be genuinely funny if you cannot be transgressive.