I can only point to an interesting dichotomy:
AI models are named as such for being descriptive models of the real world.
But then humans seem bound to use them as prescriptive (or role) models.
And so the meaning of the word model is doubled as both the representation of reality and of the reality to aspire to.
This immediate duality highlights a dangerous feedback loop: technology initially built to mirror human behavior that inadvertently shapes future behavior.
And this is oddly immediately contained in the meaning of the word “model”, used so frequently today in the context of AI that we forget the potency behind it.