My previous blog posts on AI safety and effectiveness have tended to take the perspective of those who build AI systems, since until recently users traditionally held very little responsibility in…