When Doctors and AI Interact: on Human Responsibility for Artificial Risks