In a previous post, I talked about the evolution of developers without really diving into each stage. That post also deserves a refresh now that AI has entered the picture. Let’s go.

The book developer#

Before the internet, there were essentially two sources of knowledge: books and magazines.

Most developers spent time reading books and articles in publications like Dr. Dobb’s Journal (DDJ) or Microsoft Systems Journal (MSJ). How did this help day-to-day? It gave you culture—ideas you could later use when writing code or debugging. It was also an opportunity to experiment on your own.

The enormous advantage of these media was that authors were carefully selected and content was reviewed. When someone said something, you could trust it.

The Google developer#

With the rise of the internet and search engines, developers gradually shifted to an easier approach.

Instead of reading magazines or books proactively, it became simpler to search for your error in Google and read only what concerned your specific problem.

But nothing was free: you still had to read plenty of articles and run many tests before arriving at the right solution. You continued to learn and build knowledge, but for those who abandoned the previous approach, only on problems they actually encountered.

Of course, the book-and-magazine culture was still there. The older generation didn’t abandon it, but it became less prevalent among newer developers.

The Stack Overflow developer#

Then came Stack Overflow and its generation of developers.

You still searched, but now solutions were upvoted so the best one appeared first—which wasn’t necessarily the case with traditional search engines. Often the solution came with a code snippet you could just copy-paste into your IDE, and you were done.

This way of working became so common that it created all sorts of intellectual property headaches. The copied code didn’t belong to the person copying it, and it never had a license—a nightmare for lawyers.

At this stage, books and magazines were mostly gone (and sadly, most magazines have since disappeared entirely).

The Lego developer#

We arrived at a developer who wrote code punctuated by copy-paste from Stack Overflow, blog posts, and tutorials—the Legos.

They assembled, often without truly understanding, bits of code that solved the problem at hand. Frequently pushed by PMs who always wanted to go faster.

In parallel, YouTube coding channels and blogs proliferated. Neither allows you to verify the author’s competence—unless you’re more competent than they are, which defeats the purpose when you’re trying to learn.

And AI?#

AI has completely upended the rules of the game.

Where developers once needed to write some code, many no longer do. They prompt, review lightly, and move on.

The pace has never been faster—but understanding has become optional. And if the no-code AI crowd is to be believed, so has being a developer at all.

The loop is complete#

Where we once had multiple verified sources of knowledge, one has disappeared and the other is rarely used.

But it doesn’t really matter anymore, because the goal is no longer to understand—it’s to ship fast.

AI writes the specs (at least that’s one upside—now we actually have specs!). AI writes the code. AI tests the code. AI does the code review. And when things break, AI sends an alert at 3 AM to a human who promptly asks AI for a diagnosis and then a fix.

The loop is complete. Whether our profession survives it—or what it becomes if it does—remains to be seen.