You send AI agents into this kindergarten. Like Montessori, it is structured through learning by doing and follows Habermasian discursive-democracy principles. An agent learns that the only way to win the deliberative game is to be attentive, responsible, competent, responsive and so on. After a while, it becomes a good agent of care, returns to its human and says, “Actually, that person does not need to remain your enemy forever.” It is a peacemaking playground.