Excitement
I wanted to share my excitement of forcing myself to build without using AI.
I would say I've been learning to code semi-con...
For further actions, you may consider blocking this person and/or reporting abuse
This one landed at exactly the right time for me. 😄
I've been thinking about almost the same problem recently. I use AI heavily in my engineering work, and it is incredibly powerful for building, exploring architectures and moving quickly, but I realized that speed can hide a very different question: could I still reconstruct the reasoning myself?
So I've started deliberately separating "building mode" from "learning mode". In learning mode I want the friction back: documentation, writing the code, getting stuck, debugging, and understanding why something works before asking AI to solve it for me.
I really like that you also distinguish this from simply searching the web. Looking something up still leaves the assembly of the solution in your hands. Having an AI produce the whole path can remove exactly the struggle that makes the concept stick.
And yes, that moment when something finally clicks after fighting with it for a while is completely different. 😂
I don't think the lesson is to stop using AI. For me it is becoming more about knowing when AI is accelerating work and when it is accidentally accelerating past the learning.
Really enjoyed this, Anna. 👏
I really relate to the part about getting stuck and then finally having that “oh, I get it now” moment. That feeling is easy to lose when the solution is available instantly.
I don't think AI needs to disappear completely either. For me, the useful distinction is whether I'm trying to learn something or just get something done. If I'm learning, struggling with the problem for a while is probably more valuable than getting a perfect solution in seconds.
And the nested menu is a nice example of that. It's not complicated code, but actually building it yourself gives you a much better feel for how the pieces fit together.
Exactly. I think “learn something” vs “get something done” is probably the cleanest way to decide how much AI should be involved.
When I’m building, removing friction is usually a feature. When I’m learning, some of that same friction is actually the exercise.
That nested menu example captures it well. The value is not really the menu itself, it is having to reason through the structure, state and interactions until the pieces make sense in your head.
I’m starting to think of it almost as choosing where I want the friction to live instead of simply trying to eliminate it everywhere. 😄
😂
I agree! You learn more about what can and cannot be done when you try to figure it out for yourself. Personally, I know that when I reach the point of cursing to myself in my head, a real wall has been found. You don't forget what you learned after that. 😂
I remember when I was first learning web development seeing the rails routing error page and the "stack too deep" error. I was like oooooooh! Thats why they call it stack overflow. Hahahahahaha
Thank you for your perspective! I like that you have created space with learning mode to keep your skills sharp.
I had fun reading your article and I completely agree. Sometimes doing even the basic projects without AI at all reinforces us in a way that makes more dependable in the long run. Build small but foundational concepts, when things get big, the only thing you'll just have to learn is how to connect them.
Whats wild is I can relate to this now not just from a code perspective, but from a project perspective. Ive have several apps/services/mcps connected for things im like, wow, I connected all of that and im not totally confused. Amazing! Haha.
A natural confidence booster indeed 💯
Same thing I did a week ago! I went through all fundamentals and told myself we never going to use AI. Now I am writing code myself and figuring out things when Im stuck by googling or reading docs !
I started to feel the “Yes, that is the solution” again!
Such a great, human-needed feeling!
AI has brought immense speed and abstraction to programming, leading many people to overlook complexities—the very things that make the learning journey and the day-to-day practice of coding both challenging and rewarding, whether you are creating simple things or complex systems that can be used to make a real-world impact.
Even with the AI boom, I still write all my code completely by hand, which gives me the freedom to structure my overall code and the entire project architecture my own way. Right now, I use AI mainly as a fast, intelligent documentation lookup—and actually a bit more than that, as a tool to help me interpret errors and suggest fixes, without taking away my ability to structure and solve the core problem myself.
I intend to rely more heavily on it in the future, but only once I've built a moderately solid foundation and gained enough hands-on experience developing solutions by hand.
I think the only way to grow is to actually sit with the code and think about what its doing. I couldnt imagine someone taking on a senior developer position one day, or any developer position really, having done nothing but vibe code up to that point. There is a vastly widening knowledge gap in my opinion. They say ai writes all the basic code now, well in order to get to that higher level, you have to know how code works and what it does and what its limitations are. Else, impending doom. There is alot to say about this, and I admire your passion.