essay
Part 4: How to Use AI as a Mirror to Map Your Own Judgment
Close your eyes. Touch your nose.
You didn’t miss.
There is a sense that did that. You don’t think about it. It just works. Your body knows where it is in space, even when your eyes are closed.
There is an equivalent sense for your expertise.
You don’t think about that one either. It tells you, in any given moment, what you actually know — versus what you’re pattern-matching from adjacent territory. What you have a real view on, versus what you’re improvising in confident language.
Most experts have this sense without naming it. They feel the difference.
“On this question I have a real view. On that one I’m winging it.”
That metacognition separates the practitioners you trust from the ones who sound just as confident on every topic.
AI use mutes this sense.
There Are Three Types of Operators When It Comes to Self-Knowledge
There are operators who know exactly where their judgment ends.
There are operators who think they know.
And there are operators with flat confidence across the entire territory.
Only the first one is safe. The other two are walking around with a map that doesn’t match the terrain.
This article is about how to redraw the map. Deliberately. In one afternoon.
OPERATOR FILE #11
Expert operators know AI atrophies the felt sense of where their judgment ends.
Average operators trust the felt sense even after it’s been muted.
Commodity operators never had the sense calibrated in the first place.
When AI produces fluent output for you, the natural “I don’t actually know this” signal that would normally fire when you’re outside your judgment territory — gets quieter.
The output sounds right. Your editorial pass approves it. The signal that would have said “wait, I’m guessing here” doesn’t fire — because you didn’t have to guess. The model did.
Repeat for a few hundred prompts. The sense atrophies.
You lose the topology of your own judgment. The map that tells you where your real expertise ends and your borrowed fluency begins.
Here is the useful part.
AI can also be turned around to do the opposite. To rebuild that map for you. Deliberately. In an afternoon.
The Mirror Exercise
Three steps. None of them are exotic.
Step 1 — Pick five specific claims you’d defend if challenged.
Not generic claims. Specific ones.
“In our market, the right pricing model is X for these three reasons.”
“The framework that fits this client situation is Y, not Z.”
“The best approach to this technical problem is A, and B is a common mistake.”
Five real positions. Things you would defend in a meeting on Monday.
Step 2 — Ask AI to produce the strongest counter-argument to each.
Not a hedge. Not a balanced view. The strongest available case against what you said.
Make the prompt explicit:
“Steel-man the opposing position. Assume someone smart and well-informed disagrees with me. What’s their best argument?”
The model is good at this. Let it run.
Step 3 — Read each counter-argument and notice your reaction.
Two reactions matter. Watch for them precisely.
The Two Reactions
The first reaction is: “this is wrong, and here’s specifically why.”
The counter-argument doesn’t land. You can name the flaw in two seconds — the assumption it makes, the variable it ignores, the case it fails on.
The judgment fires.
That tells you where your judgment is real.
The second reaction is: “huh, that’s actually a good point — I hadn’t considered that.”
Or: “I’d want to think about that more.”
Or — the most dangerous one — silence. You read the counter-argument and have nothing specific to say back.
That tells you where your borrowed fluency begins.
OPERATOR FILE #12
Expert operators know their territory is not uniformly mapped.
Average operators assume they’re equally strong across the whole field.
Commodity operators have flat confidence and don’t know it’s a problem.
Most experts assume their territory is uniformly mapped.
Run this exercise once and it isn’t.
Most knowledge entrepreneurs, doing this exercise honestly, find that maybe two of their five “defended” claims hold up cleanly. The other three reveal soft spots — places where they’ve been speaking with the confidence of expertise but actually drawing on borrowed scaffolding.
The fluency was real. The conviction underneath was not.
This is not bad news.
This is a map.
Once you see the soft spots, two things happen.
The territory you actually own becomes clearer. Your confidence in those claims sharpens, because you tested them and they held.
The territory you don’t own becomes negotiable. You now know which prompts you can trust your gut on. And which require deferring to someone whose gut is more developed.
That distinction is what separates operators who use AI well from operators who use it confidently.
The map is the difference.
The Quieter Daily Version
Once you’ve done the formal version, there is a faster version you run in your head.
When you’re in a real situation — a client conversation, a strategic decision, a technical call — and you’re about to take a confident position:
“If someone smart pushed back here, would I have a specific answer? Or would I have a feeling?”
If it’s a specific answer: ship the position.
If it’s a feeling: slow down.
The feeling might be right. It also might be borrowed.
The thirty seconds it takes to notice the difference is the cheapest insurance you’ll buy this quarter.
OPERATOR FILE #13
Expert operators know the felt sense lies. The mirror tells the truth.
Average operators trust the felt sense and assume their map is current.
Commodity operators don’t run any check at all.
The mirror works because AI is exquisitely good at producing the borrowed-fluency version of any position. Including positions that disagree with yours.
When you read AI-generated counter-arguments and have nothing to push back with, you’re reading the mirror image of your own borrowed fluency.
The model can produce it because the territory is already commodity.
You can’t push back specifically because you never built specific judgment in that territory.
The mirror is honest in a way the felt sense often isn’t.
The felt sense will tell you what you want to hear.
The mirror tells you where the work is.
Run it once on five claims. Run it once a quarter on the claims you’ve added since.
The map is small. It also compounds.
The experts who know precisely where their judgment ends are the ones who can keep extending it. Deliberately. In the directions that matter.
The ones who don’t have the map keep speaking with the confidence of expertise across the full territory.
For a year or two, both look identical.
After that, the difference becomes the only thing the market is paying for.
Declaration
You are a judgment-builder. You protect what you’ve built. You do the slow work.
Now add this.
Put your hand on your heart and say — out loud:
“I know where my real judgment ends. I don’t pretend beyond that line.”
Say it again. Louder.
“I know where my real judgment ends. I don’t pretend beyond that line.”
That is the topology discipline. Stand in it.
Action
This afternoon, do this one thing.
Write down five specific claims you would defend in a meeting Monday. Real, specific, positions.
Open AI. Ask it to steel-man each one — the strongest counter-argument it can produce.
Read them.
Notice which ones you can refute in two seconds, and which ones leave you silent.
The silences are your map.
That map is worth more than the confidence you walked in with.