Learners vs. Yearners

AI pulls engineers in two directions at once.

One is excitement. A new model every week, a new tool, something you can ship in an afternoon that used to take a month. If you like learning, there has never been a better time to be an engineer. Call these people learners.

The other is longing for the puzzle: the afternoon lost to one bug, the moment a hard problem finally gives, the quiet satisfaction of a system you understand all the way down. More of that work goes to agents every month. Call these people yearners.

People talk about them as two camps: the ones who embrace AI and the ones who resist it. They’re not two camps. Most good engineers I know are both, often on the same day. I am. You probably are too. You use the new tools every day, and you still miss [insert the kind of problem you used to love solving].

Each impulse alone has a failure mode.

The pure learner chases every new tool and never stays long enough to get deep. They can make an agent produce a lot of code. They can’t always tell you whether it’s good. When it breaks in a way the agent can’t fix, they’re stuck, because they never built the mental model underneath.

The pure yearner has the opposite problem. They know exactly what good looks like, and they refuse the tools that would let them build ten times more of it. The craft becomes something to protect instead of something to use. That’s how skilled people end up left behind, not because they lacked ability but because they mistook the method for the point.

Because the point was never typing. What the puzzles gave us was judgment. Every hour spent chasing a bug was an hour spent building a model of how systems actually behave. That model is what lets you look at an agent’s work and feel that something is off before you can say why.

That’s why the yearning matters. You can only miss the craft if you had it. The longing is evidence of taste, and taste is what the tools can’t supply. An agent will climb whatever hill you point it at. Someone has to know which hill is worth climbing.

Which raises the hard question for anyone starting out now. If agents do the puzzles, how do you get something worth yearning for?

I think you have to choose some of the puzzles on purpose. Build one thing the slow way. Read the code the agent wrote until you understand it, not just until it passes. Debug one problem yourself before you hand it off. It will feel inefficient. It is inefficient, in the way practice always is. But it’s how you earn the judgment that makes everything else work.

So don’t pick a side. Learn the new tools like a beginner, and keep the standards of someone who did it the hard way. If you’re early in your career, go out of your way to earn those standards.

Those are the engineers I want to work with.