Do we agree it's a bad sign when a model treats lived human experience as noise?