My problem is with the dataset and datasets like this overall that sets the tone through AI of what is acceptable and what is not.