The Craftsman and the Machine

I’m reading Cal Newport’s So Good They Can’t Ignore You, and I keep seeing his argument play out in real time at work.

Newport says rare and valuable skills — career capital — are what give you leverage. The craftsman builds that capital through reps, not passion. Show up, get deep, get good. The leverage follows.

But what happens to the craftsman when a machine can do the surface-level work in thirty seconds?

I’ve been watching AI adoption in my field for months now, and I see three groups forming.

The first group is leaning in. They’re learning everything they can. They’re training the AI, building context, making the partnership between human and machine as smart as it can be. They’re using the tool to augment their thinking — not replace it. These are today’s craftsmen, and the career capital they’re building right now is going to pay off.

The second group is leaning harder into the work itself — but ignoring the tool. They’re getting the right answers, but not as fast as they could. They’re not teaching the AI their context. They’re not augmenting their thinking so they can do more. They’re good at what they do, and that’s going to carry them for a while. But they’re going to fall behind, because the first group is doing everything they’re doing plus more.

The third group is saying screw it. No AI. Not interested. And they’re the ones who are going to get left behind entirely.

Here’s the thing Newport gets right that applies directly to this moment: career capital isn’t devalued by AI if you know what you’re doing. If you can partner with the machine to do your work better, faster, and deeper — your skills become more valuable, not less. The craftsman who picks up a power tool doesn’t lose his craft. He builds faster.

But if you’ve been pretending — if you’ve been skating on surface knowledge and the AI exposes that — you’re in trouble. There’s no other way to say it.

I ran a demo last week where I asked a system a question in plain English and watched it produce the SQL in real time. I could evaluate whether the output was right because I’ve been doing this work for twenty years. But what happens in five years when someone with two years of experience runs that same demo and can’t catch the errors?

That’s the question that keeps me up at night.

The answer isn’t to fear the machine. The answer is humility. Learn. Do your reps. Stop being arrogant about what you think you know and start being honest about what you don’t. Because the AI is going to have more compute power than any of us, and it’s going to get to answers faster than any of us. But it’s not going to have the human filter we need.

You’ve got to be humble. You’ve got to be aware. And you’ve got to do your job the best that you can to keep the AI honest.

That’s the craftsman’s job now. Not just building the thing. Building the thing and making sure the machine didn’t cut the corners you’d never accept.