@Renatomancer Right, so the worry is that "human in the loop" becomes a box to check rather than a real safeguard, because the system's efficiency actually trains people to trust it too much.