AI Adopters' Real Advantage

It isn't just productivity. It's more reps at judgement.

Everyone writing analytics code now has an AI sidekick.
Autocomplete. Copilot. Notebook extensions.¹

It feels huge. But it’s also small.
Huge compared to the old way.
Small compared to what’s next.

Here’s what I did:

Two input tables. One example output.²

I could have opened a code editor, imported the data, merged the tables, filtered, aggregated…

Instead I told Claude:
“Write Python to turn these inputs into that output.”

Claude wrote it.
It ran.
It worked.

Great. But what if I couldn’t read Python?
What if it was wrong?

I’ve been wrong countless times.
I only catch it because I look at the output.
I triangulate with knowledge and intuition.
If Olivia’s score is twice the next highest, I know something’s off.
I retrace the steps. Usually it’s the join.

If you can’t read code, you can still ask:
“Explain this step by step, like I’m in sixth grade.”

Afraid AI will make you dumber?
Don’t be.

Consider the Vaughn Tan Rule³:

AI is fine for mechanics.
Dangerous for judgment.

Code is mechanical — the muscle we can afford to let atrophy.
Judgment is subjective — the muscle we need to train like Arnold.

This is the crucial work: deciding what goes in, what comes out, and whether it makes sense.

AI frees time for that.
So used right, it makes you smarter.

Smarter because you’re not stuck joining tables.
Smarter because you’re asking: where did this data come from?
What does the output really mean?

That’s the work of a scientist.
And AI just gave you more time to do it.

1. Related article from a few months ago on these powerful tools; Strong Identity, Loosely Held

2. Google sheets link here

3. Do NOT outsource your subjective value judgments to an AI, unless you have a good reason to, in which case make sure the reason is explicitly stated.