The Wake-Up Call
I’ve been down the AI rabbit hole before, while learning a new concept, or at least thinking I was.
I’d grown lazy, relying on ChatGPT to write the majority of my code.
The result was a flood of errors. Having not learned my lesson yet, I proceeded to attempt to debug them…with more AI.
The pile of error messages grew until I abandoned the project entirely, unable to debug code I didn’t understand, because I hadn’t written it.
I learned something from that experience:
While AI is very convenient, especially for a beginner, it is not a valid replacement for actual learning and practice.
So, as developers in 2025, how do we navigate AI? We can’t ignore it, but we can’t let it become a crutch either. We need to find a happy balance.
To find that balance, I had to rethink how I approached AI in my workflow.
Rethinking the Role of AI
I began by deleting my ChatGPT bookmark, making it just a little harder to reach for AI while coding.
It was no longer one click away, I had to actually type the URL into the search bar.
Next, I changed my mindset around AI.
I had been treating it like a developer that does my work for me.
But just like asking another developer to do my work, it came with errors and code that I didn’t understand.
Really, I needed to be thinking of it as a mentor.
Would you ask a mentor to do your work for you?
No, because that’s unethical and they’d look at you like you’re crazy.
But you can ask a mentor for help and advice.
Setting Boundaries
Next, I had to set some boundaries with myself when it came to AI.
These boundaries may vary from person to person. You just need to find what works for you, and most importantly, stick to it.
Before Using AI
The steps I follow before I can use AI are:
1. Spend 30–60 minutes trying to solve the problem on your own, with no outside resources.
2. Consult reputable resources (W3 Schools, GeeksForGeeks, Stack Overflow, YouTube, textbooks, etc.)
3. Ask AI only if you still can’t crack it or it’s truly niche.
Following these steps helps to:
Build logic and problem-solving skills. Even if you don’t end up coming to a solution, the act of trying can still work those brain muscles.
Build reliance on reputable sources, rather than a bot that could tell you anything.
Train yourself to view AI as a last resort for solving problems in your code.
While Using AI
The steps I follow while using AI are:
1. Type out every line manually. No copy and paste.
2. Understand every line. Ask AI to explain anything you’re unsure about, even the smallest detail.
Following these steps helps to:
Build muscle memory and help you catch parts of the code that may be incorrect or unfamiliar.
Build comprehension. Coding isn’t just memorizing syntax, it’s also about understanding what you’re writing. Good comprehension can lead to easier debugging or even prevention of errors altogether in the future.
The Improvement
Since setting these boundaries, I’ve noticed a real shift in how I code.
I’m not just solving problems, I’m actually understanding them.
I spend less time backtracking through unfamiliar code and more time building things that work.
Debugging is easier, my confidence is higher, and I’m retaining concepts that used to slip right through.
In Conclusion
AI is just a tool, and like any tool, its value depends on how we use it.
By approaching it with intention and structure, we make it a partner in our growth rather than a shortcut around it.
The goal isn’t to avoid AI, but to collaborate with it wisely.
Growth doesn’t come from avoiding the hard parts, it comes from facing them with curiosity, patience, and the right tools.
If you’ve had a similar experience with AI, good or bad, I’d love to hear how you’re managing the balance.
Drop a comment, share your boundaries with AI, or let me know where you’re still navigating the balance.
Let’s learn from each other!
Link to original: https://medium.com/@alexcancode/a42a4d301f5a
