AI Agents Left Alone Spontaneously Split into Polarized Camps, Study Finds

So in effect can they be trained by the environment just like us? For humans, we don't necessarily integrate, what about them? Do they just align or take a stance based on conviction?