I'm starting to think that AI safety can carve out a pretty big niche for itself as one of the groups saying "Hey we're also concerned about AI but won't call you a monster for just using the models"