Imagining Trustworthy Autonomous Systems