In the early 1980s, I wrote my first lines of code in C. There were no tutorials, no Stack Overflow, no AI tutors—just a reference book by Kernighan and Ritchie, a blinking cursor, and a lot of trial and error.
Then came Lint.
For the uninitiated, Lint was a static code analyzer—a glorified tattletale that flagged your sloppy habits, scolded you for undeclared variables, and explained why you shouldn’t do what you just did. And after being called out 5, 10, 20 times for the same thing, something fascinating happened:
You stopped making the same mistake.
Not because someone told you to.
But because your brain re-wired.
Fast forward 40 years.
Today, instead of Lint for code, we have LLMs like ChatGPT for thought.
Same principles. Different battleground.
🚨 Modern Lint: ChatGPT for Mental Models
When I use ChatGPT as an AI assistant—especially during productive hours—I’m not just offloading work. I’m getting feedback in real time:
- “This paragraph is confusing. Try this instead.”
- “Your logic jumps here. Want to bridge it?”
- “That term’s not accurate. Here’s a better one.”
Every correction is a micro-lesson. Every suggestion is a recalibration.
The first time, you nod.
The fifth time, you sigh.
The tenth time, you stop making the mistake.
Just like Lint taught C programmers to avoid buffer overflows,
ChatGPT teaches professionals to avoid idea overflows.
🧪 The Subtle Rewire: Evidence It’s Working
You might think, “Well, that’s just anecdotal.”
Fair. But here’s the kicker: research is starting to catch up.
📊 Early Evidence:
- Wharton MBA (Ethan Mollick):
Students who used ChatGPT wrote better papers and learned to write better even when not using it. - Stanford + GitHub Copilot Study:
Developers using Copilot produced code faster, made fewer errors, and over time showed signs of adopting better coding patterns naturally. - LLM Tutoring Studies (Google & OpenAI):
Users who interacted with LLMs in tutoring modes self-corrected more over time, and retained more knowledge than traditional methods.
It’s not just that LLMs help you now.
They help you become someone who needs less help later.
🕒 When the AI Is Down (3–7pm Club)
Let’s be honest: there are hours of the day (usually between 3–7pm) when ChatGPT gets buggy or gives worse responses. It’s like watching a caffeinated intern crash after too many scones.
But here’s the thing—
When the tool gets worse, you’ve already gotten better.
By that time, your habits have improved:
- Your writing is clearer.
- Your thinking is more structured.
- Your analysis is tighter.
You don’t flounder. You fly on autopilot—at least until it comes back online.
🎯 Will You Need AI Less Over Time?
In a way, yes.
But not because AI gets worse—because you get better.
Eventually, your relationship with your assistant shifts:
| Early Stage | Later Stage |
|---|---|
| “Correct me” | “Challenge me” |
| “Fix my logic” | “Pressure test my logic” |
| “Write this for me” | “Help me think this through” |
| “Give me ideas” | “Refine my best idea” |
You go from student → partner → peer.
🧭 The Takeaway: Why AI Is the New Lint
We didn’t stop writing code when Lint showed up—we just wrote better code.
Similarly, we won’t stop thinking, writing, or leading when AI assistants are available.
We’ll just do all of it better.
And if you’re still skeptical, I’d love to see long-term studies showing how performance changes after 3, 6, or 9 months of AI use.
Until then, I’ll keep asking hard questions, building with Microsoft tools, and—yes—occasionally having a beer in the forest with a raccoon named Doug.
Because even in the woods, progress marches on.
💡 P.S. Want to Test Yourself?
I’m developing a free self-assessment checklist to track how AI improves performance over time.
Think of it as your personal mental Lint log.
Let me know if you’d like a copy.
Want to learn more? Check out our hub of information.