A lot of the biggest AI hypers read "a computer cannot be held accountable" and thought "cool, I'll tell the computer to make all of my decisions and escape responsibility" without reading the second part.