What to learn in the age of AI
↩ Follow-up to The skill that survives AI
After the last post, the obvious next question: if AI keeps eating layers of competence, what should I actually learn to stay competitive?
My answer is a filter, not a list. Ask about any skill: is this a layer AI eats first, or a layer it reaches last? Then only invest identity in the second kind.
Not worth deep investment: frameworks, language syntax, tooling. Use them, obviously — just don't build your professional identity on them. This is exactly the layer AI takes first, and every hour spent memorizing an API surface is an hour with a shrinking half-life. It feels productive because it's measurable. That's a trap.
Worth investing in — four things:
Judgment under uncertainty. Deciding with real stakes and incomplete information, then living with the outcome. No course teaches this; your own projects do. Every "kill it or keep going" decision is a rep. Basic decision economics helps as a foundation — opportunity cost, expected value — plus knowing your own cognitive traps well enough to catch them mid-decision.
Reading people and markets. AI computes X and Y; it doesn't feel what a customer is actually buying, when a partner is bluffing, why a team quietly resists a decision. Sales, negotiations, live conversations with users. Muscle, not theory.
Taste. When generation is free, telling good from mediocre becomes the scarce skill — in products, code, text, design. Taste is how you accept AI's work once you stop reading every line. It's built on volume: study things that worked and ask why this choice and not another.
Depth in one thing. Counterintuitive, but a pure generalist loses to AI, because AI is the perfect generalist. A person who knows one area deeper than the model sees where the model is confidently wrong. Whatever your fifteen years taught you — keep that alive instead of writing it off.
The common thread: AI automates the middle of the stack — production, syntax, execution. What's left is the top (what's worth doing and why) and the bottom (is this actually good). Learn toward the edges.