Can AI not feel bad because of a mistake?