On why learning slowly is still valuable in the age of shipping fast.
I'm currently learning a developer tool that's new to me and I'm deliberately doing it the old fashioned way. Reading a programming book and hand-typing the code examples. Getting reps in writing and executing the syntax. Debugging the errors I make. Growing an understanding of how it all works.
Page by page.
Brick by brick.
Building my personal context.
This sounds—and admittedly feels—prehistoric in the age of AI. It's taking a lot of resolve to not just close everything, rip open Cursor and just prompt something into existence. Truth is, that's always an option. I could be doing it now instead of writing this blog post. But what will it get me?
I'll create a working app which I don't understand and cannot explain.
My experience with AI generated apps is this:
And most worryingly for the latter, the end is result is never one that I understand.
With AI I've built video processing apps, audio editing apps, React Native apps—things I could not have done before. But I didn't get any of them to completion. I couldn't tell you how they worked. I don't know if they were made very well.
I gained no additional skills in the process. I didn't learn anything.
Because I didn't make them, AI did.
Sure we can ship faster than ever before. Awesome. But the ability of any of us to increase our knowledge is still rate-limited.
So I'm choosing to learn slower to understand better. To build up the personal context I'm going to need to create and explain concepts going forward.
To inform the prompts I'll soon write to create at the speed of thought.
Books first, prompts later.
