Lately, I’ve been thinking about how we perceive AI code generation.

We use words like revolution, disruption, replacement. And sure, it’s impressive. But when I zoom out, it feels less like a revolution and more like… version 5.0 of autocomplete.

And honestly, that framing has changed how I think about it.

Writing Has Always Evolved

If you take the long view, writing has been on a steady trajectory:

  • Handwriting
  • Typewriters
  • Word processors
  • Spellcheck
  • Grammar correction
  • Autocomplete

Code followed the same path:

  • Plain text editors
  • Syntax highlighting
  • Real-time linting
  • IntelliSense
  • Snippet completion

Tools like Visual Studio Code and IntelliJ IDEA already complete half my thoughts. I type three letters, and the IDE guesses the rest. It fixes imports. It warns about types. It underlines mistakes before I even think about running the program.

We’ve been collaborating with our editors for years.

Now tools like GitHub Copilot just complete more.

From Syntax Completion to Intent Completion

Traditional autocomplete works at the token level:

user.get_na  # → get_name()

GenAI works at the intent level:

# create a function that validates JWT and fetches user profile

And it generates something usable.

That jump feels dramatic, but technically it’s just a shift in abstraction. Instead of predicting the next word, it predicts the next structural pattern.

It’s still completion.

Just… zoomed out.

The mind-shift...

What’s changed for me is this: I no longer treat GenAI as something separate from my development flow.

I treat it like:

  • Boilerplate autocomplete
  • Refactoring autocomplete
  • Test-writing autocomplete
  • “I don’t feel like typing this again” autocomplete

It’s not replacing design thinking. It’s compressing expression.

When I describe what I want and let it scaffold the first draft, I’m doing the same thing I’ve done for years with smaller completions—just at a larger scale.

And I still review everything the same way I’d review a PR. Because in a way, that’s what it is: a fast, slightly overconfident pair programmer.

Some final thoughts

The narrative around GenAI often makes it sound like something fundamentally new.

But when I strip away the hype, it feels like the next logical step:

Handwriting → typing → syntax awareness → autocomplete → intent-level completion

It’s not that AI suddenly started “writing code.” We’ve been letting tools help us write for decades and now the tool just understands a bit more context.

And the more I think of it as smart autocomplete—not artificial genius—the more naturally it fits into how I build software.