For a century, the career advice was the same: specialize. Go deep, own a niche, become the person who knows the most about one narrow thing. It was good advice, because knowledge was scarce and depth was how you made yourself irreplaceable.
That logic is inverting. When a model holds the specialized facts of every field and can recall them on demand, deep recall stops being a moat. The scarce skill becomes the one specialization trained out of us: the ability to connect across domains — to see how a pattern in one field solves a problem in another.
Why depth was a response to scarcity
Specialization was never the goal; it was the adaptation to a world where knowing a lot about everything was impossible. So we divided the knowledge up. Remove the scarcity — give everyone instant access to any field’s facts — and the adaptation loses its edge. The bottleneck moves from “knowing the field” to “knowing which fields to combine.”
When the machine is the specialist, the human advantage is synthesis — the connection no single field could see.
Generalists and experts, together
This isn’t the end of deep experts — it’s a change in how their depth gets used. The generalist (human or AI) ranges wide, finds the connection, and knows the precise point where they need someone who has actually gone deep. The expert supplies that depth on demand, for the specific question, and goes back to their work.
The future isn’t generalists versus specialists. It’s a generalist agent that knows when to call a specialist — and a market that lets it. The breadth finds the question; the depth answers it.