Somehow, half the year is gone already (Happy Canada Day 🇨🇦 and Happy Fourth of July 🗽). It felt like a good excuse to stop, look up from the keyboard, and take stock of where my head has drifted over the last six months.
And where it’s drifted, mostly, is AI. I’ve mostly been questioning the underneath: what is this actually going to be to us, long term? I’ve spent a lot of these months reading — papers, blog posts, and listening to those living in the AI space who have watched it up close. And I have come to the conclusion that it is…
Another tool on the belt
For a while I thought of AI the way I think of everything else I’ve picked up as a developer: another tool in the arsenal. Language syntax was a tool. Design patterns were a tool. The IDE was a tool. Each one made me a little faster, a little sharper, a little more capable than I was without it.
And I think AI sits there too. It’s on the tool belt, right next to the rest of it.
But the more I thought about it, the more I felt like that framing undersells what’s happening. AI isn’t just a tool on the belt. It’s starting to look more like the belt itself (I know, I know, I can see the eye rolls from here). But it’s starting to feel like the thing that holds all the other tools together and hands them to you when you reach for them. It’s not just a hammer. It’s the thing that lets you carry every hammer you’ve ever owned at once (or hoped to own), and pull out the right one before you’ve finished the thought.
Personally, I think that’s a bigger deal than “faster autocomplete.” A tool makes one task easier. A tool belt changes how you carry yourself through all of them.
But a belt is only as good as the pants
But for all the credence I’m giving it, it’s worth looking at its shortcomings too.
A belt doesn’t do anything on its own. It only works because there are pants for it to hold up. You can have the finest belt in the world, but if you’re not wearing any pants underneath, all that belt is doing is drawing attention to the problem.
The pants are The Fundamentals. The languages. The concepts. The knowledge. Knowing why a thing works, not just that it does. That’s what the belt cinches around. AI can hand you an answer, but you’re the one who has to know whether it’s the right answer, whether it fits, whether it’s about to fall apart under load. If you don’t understand what’s underneath, the belt has nothing to hold onto.
So no, I don’t think AI lets us stop learning the deep stuff. If anything it raises the stakes. The people who lean hardest on the belt while skipping the fundamentals are the ones most likely to end up exposed. And the last thing any of us wants is to be caught, one day, with our pants on the ground. 😅
Keep your pants on. Learn the fundamentals. Then the belt is worth something.
Learn to write, and write well
The other thing I want to say a bit louder: learn to write. And then keep going until you write well.
I’ve been convinced of this for a while now (it’s half the reason this blog exists), but the AI shift has turned it from a nice-to-have into something closer to a core skill. Development is moving more and more toward natural language. The interface between what you want and what gets built is increasingly just… words. And the better your words, the better everything downstream of them gets.
Remember, these models are next-word predictors. Which means the context you provide is the raw material for everything they produce. Vague, thin, sloppy input gives you vague, thin, sloppy output (no tool will help you escape “Garbage In, Garbage Out”). Precise, rich, well-structured input gives you something you can actually use.
And I strongly believe this multiplies on itself. The more you read and the more you write, the more vocabulary you pick up. The more vocabulary you have, the more precisely you can say what you mean. The more precisely you can say what you mean, the better the context you can feed the model. And the better the context, the better the output.
Writing well isn’t a soft skill sitting off to the side anymore. It’s upstream of the quality of everything you build with these tools. Getting good at it is one of the highest-leverage things you can do.
Halfway thoughts
So that’s where I’ve landed at the midpoint of the year. AI is becoming the belt, if it isn’t already. I think that is exciting. It also means the two least fashionable things in tech, the fundamentals and the ability to write a clear sentence, are the very skills we must become more proficient in, not less. Keep your pants on, and learn to describe them well.
Here’s to the second half. See you in the next one — who knows what I’ll be thinking about by then?