How do we trust a human-like technology that we can't understand, that makes mistakes, and that might take our jobs?