A Personal Reflection

Let me start with a confession.

For the past two years, I've been watching the AI revolution unfold with the same mix of fascination and dread as everyone else. I've read the doomsday headlines. I've seen the demo videos that make my stomach drop. And like most people, I've tried to make sense of where I fit into this new world.

After a lot of late-night thinking, I landed on a summary that felt right to me:

"If you are a classical implementer or executor, a traditional writer or coder, a conventional engineer or normal programmer, AI is coming for you. But if you are a true thinker, real problem solver with deep expertise in a certain domain, then AI is your massive leverage."

I liked this framing. It drew a clean line between the people who should be worried and the people who would be fine. But the more I've watched AI evolve, the more I've realized this picture is incomplete. In some places, it's just flat-out wrong. And if we're going to navigate this shift without losing our minds—or our livelihoods—we need a better map.

---

The Problem With Drawing Lines

Let's be honest: the "thinker vs. doer" distinction feels good because it lets us sort the world into two boxes. You're either in the danger zone or you're safe. Simple.

But life isn't simple. And neither is AI.

Here's what I got wrong in my original thinking: I assumed that "deep expertise" was a fortress—something AI couldn't breach because it lacks true understanding. And in the short term, that's absolutely right. AI doesn't know anything. It's a statistical parrot, not a mind.

But here's the uncomfortable truth I've had to sit with: AI doesn't need to understand to be useful. It just needs to be good enough.

And "good enough" is getting scary good.

I watched a colleague—a brilliant lawyer with twenty years of experience—feed a complex contractual dispute into an AI and get back a legal analysis that would have taken a junior associate three days to produce. Was it perfect? No. But it was 80% of the way there in thirty seconds. And that 20% gap? That's closing faster than any of us want to admit.

So no, the "true thinker" isn't automatically safe. Not if their expertise lives in textbooks, case studies, or recorded decisions. AI has read all of those. Twice.

---

The Real Danger Isn't Replacement. It's Complacency.

I think we've been asking the wrong question.

We keep asking, "Will AI replace me?" But that's like asking, "Will cars replace horses?" The answer isn't yes or no. It's: "What happens to the people who refuse to get in the car?"

The classical coder who insists on writing every line from scratch? They're in trouble. Not because they're a "doer," but because they're ignoring a tool that makes them ten times more productive. The traditional writer who refuses to use AI for research and outlining? They're not being noble. They're being stubborn.

But here's the twist I didn't see coming: the same AI that threatens the "doer" can transform them into a "thinker."

I know a programmer who was the definition of a "normal coder." He took tickets, wrote functions, fixed bugs. Solid, reliable, replaceable.

Then he started using AI as his pair-programmer. He'd prompt it for the boilerplate, review the output, stress-test the edge cases, and integrate the pieces. He didn't write less code; he wrote more—but at a higher level. Suddenly, He had time to think about architecture. He started asking why features were being built, not just how. He became the person management went to for design decisions.

He didn't get replaced. He instead got elevated. Because he stopped acting like a machine and started acting like a human who uses machines.

---

What I've Learned About Expertise

Here's the honest truth about deep expertise, and it's a hard pill to swallow:

Expertise only protects you if it's visceral.

If your knowledge is the kind you can write down—step-by-step instructions, decision trees, best practices—then it's already AI's territory. You're essentially a walking manual, and manuals are cheap.

But if your expertise lives in your gut? If it comes from years of failure, from subtle patterns you can't quite articulate, from intuition sharpened by a thousand bad calls? That's different. That's the kind of knowledge that doesn't make it into training data. Because it was never written down.

I think of it like this: AI can tell you every rule of improvisational comedy. It can recite the principles of "Yes, And" and timing and character work. But put it on a stage with a live audience and a suggestion like "dentist's office," and it will freeze. Because improv isn't about rules—it's about reading the room, sensing the energy, taking risks in real time.

That's the human advantage. Not intelligence. Embodiment.

---

So Where Does That Leave Us?

Let me rewrite my original summary, now that I've had time to think it through:

"If you are a classical implementer who merely translates instructions into output, AI is coming for you. But if you are someone who defines problems, navigates ambiguity, exercises judgment, and possesses intuition that can't be reduced to data—then AI is your exoskeleton. Just don't get comfortable. The moment your expertise stops evolving, it becomes data. And data is AI's native language."

Does that feel different? It should. Because it acknowledges that this isn't a one-time reckoning. It's a continuous process.

The people who will thrive aren't the ones with the most knowledge. They're the ones who stay curious. Who treat AI as a collaborator rather than a competitor. Who use it to offload the boring stuff so they can focus on the human stuff: connection, creativity, judgment, and the messy, beautiful work of figuring out what actually matters.

---

A Simple Way to Think About Your Future

Here's the framework I've started using with myself and my friends:

• If your job is about executing clear instructions, start learning to instruct AI. Your value is shifting from doing to directing.

• If your job is about applying established knowledge, start asking yourself what you know that can't be found in a book. That's your gold.

• If your job is about navigating uncertainty, congratulations—you're already in the right place. Now just add AI to your toolkit and watch what happens.

I'm not going to tell you this is easy. It's not. It's uncomfortable to realize that the skills you've spent decades building might be commoditized in a matter of years. I feel that anxiety too.

But I also feel something else: possibility.

Because for the first time in history, we have a tool that can handle the grunt work of thinking. That frees us up to do the deeper work—the work of understanding each other, of building things that matter, of solving problems that don't have clear answers.

That's not a threat. That's an invitation.

---

The Bottom Line

Stop asking whether AI is coming for your job. It's a question that keeps you stuck in fear.

Instead, ask: What can I do today that AI can't do at all? And What can I do with AI that I could never do alone?

The answers to those questions are where your future lives.

And if you're honest with yourself, I think you'll find they're pretty exciting.