I was a bit alarmed looking at some of my recent code commits:

• Improve spacing and typography consistency across the site
• Improve inline link styling and color consistency
• Reorganize Photos section to full-width layout
• Improve mobile layout density and readability
• Fix frontend test quality and eliminate anti-patterns
• Refactor endpoint_form_controller to follow Stimulus best practices

Looks productive, right? Except I spent hours making barely visible changes to my blog and projects while avoiding the actual work I needed to do. Welcome to generative procrastination: where AI helps you avoid real work at superhuman speed.

Traditional procrastination had natural limits. Want to reorganize your desk instead of writing that PRD? Fine, but how long can you really spend arranging stuff? Want to refactor that old code instead of tackling the gnarly bug? Sure, but eventually you’d hit the tedium wall and give up.

AI has destroyed those limits.

Now I can generate infinite busywork that looks, feels, and commits like real work. I can refactor entire areas of a codebase in an afternoon. I can build elaborate CLI tools to solve problems I sparingly have. And it all happens at lightning speed, leaving a trail of commits that would make anyone feel very good.

It’s a perfect dopamine loop. Every prompt answered, every file updated, every test passing. You’re not scrolling Twitter, you’re shipping! You’re not playing video games, you’re building! Your brain can’t tell the difference between motion and progress anymore.

When you browse Reddit for three hours, you know you’ve wasted time. The guilt of traditional procrastination is immediate and useful. But when you’ve spent those same three hours pair-programming with Claude to build a comprehensive Tailwind component library for your side project? That’s different. You’ve got tangible, deployable output. But was it useful? Or was it just generative procrastination?

The most dangerous part is how it hijacks our identity as builders. We’re the people who bias toward action. AI tools let us maintain that identity while completely avoiding the hard, important work.

I’ve watched myself do this for a while now. For example, I have a customer waiting for their children’s storybook to be updated with suggested edits. It means learning how to upscale all the illustrations, recreate some illustrations, redoing the book layout, and communicating to the customer. But here I am, redesigning and refactoring my projects super fast. It’s just so easy to pair with Claude and quickly knock things out that used to take a long time!

The output is often very good. That refactored code? Actually cleaner. That documentation? Actually helpful. We’re not making garbage; we’re making high-quality irrelevance at high speed.

AI tools are powerful, but they make it even more important for us to be super honest about what’s actually nedle-moving. The strategic thinking, the ambiguous problems, the human conversations still feel like work because AI can’t instantly solve them.

Ask yourself: “Am I using AI for the well-defined, fun stuff instead of the messy, important stuff? Would I prioritize this task if I had to code it manually?” Sometimes the answer will be “I’m just enjoying building this temp side thing,” and that’s fine. We all need our outlets. Like this article itself is for me. But it isn’t work still, just new kind of recreation.

The real thinking still happens with a pen and paper, far away from AI. But good luck explaining that to your contribution graph. Anyway, I should stop writing about generative procrastination and tackle that user feedback I got for Bookling. Right after I fix this blog’s typography one more time?