An FAQ bot I built for a student poster session taught me how an AI that only wants to be helpful can end up suggesting something unethical.
