Are technologies that generate code our enemies?
The answer is no. Technology exists to help.
But the result is not always what we expect.
At the end of 2022, ChatGPT was released, one of the most impactful tools since the invention of the internet. Since then, we have seen fast changes in society and software development was no exception. We joined the wave: using AI to solve doubts, generate solutions, and mainly write code.
With this growth, something interesting started to happen. Technical forums began to change. Platforms like Stack Overflow started receiving fewer questions. The pattern shifted: fewer public doubts, more private prompts.
This shows an important transition: people stopped asking why and started asking do it for me.
For a senior developer, using a prompt to generate a class can be just a productivity shortcut.
But what about a beginner?
If even Linus Torvalds uses AI tools, why couldn’t we?
The difference is a detail that is often ignored:
does the beginner understand what the machine is writing?
This brings us to the idea of comprehension debt.
Just like technical debt comes from technical shortcuts, comprehension debt comes from mental shortcuts. It is the accumulation of code that works, but no one really understands not even the person who wrote it.
You copy, paste, and adjust… but you don’t understand why.
This creates a false feeling of progress.
The task is done, but learning did not happen.
Imagine a ship sailing without a compass or a map. It moves… but it does not know where it is going.
Coding without understanding is exactly that: producing code without knowing the reasons behind the decisions.
It is not bad code.
It is ownerless code.
AI does not create this problem.
It amplifies it.
With a single prompt, code appears and the task seems solved. This creates a sense of efficiency, but also shallow learning.
If AI solves it fast, who is learning?
How can you know if you are building comprehension debt?
Some simple questions help:
- Could you implement this feature without using prompts?
- Can you explain what the generated code is doing?
- Are you afraid to change the code without asking AI?
- Do you only test on localhost and hope it works?
If you cannot explain the solution, you probably did not understand the problem.
And if you only test and hope… something is wrong.
The cost appears over time.
You grow slower, become dependent on tools, and weaken your foundation as a developer.
In a job interview, without AI, how would you perform?
You trade speed today for stagnation tomorrow.
The goal is not to stop using AI.
It is to change how you use it.
Practical advice:
- Ask the AI to explain the code it generated.
- Rewrite the solution in your own words.
- Run the program flow in your head.
- Make small changes and see what happens.
- Try to write first and ask for AI review later.
- Use AI as a copilot, not as the driver.
As a helper tool, it is amazing.
But only if you are navigating with a map and a compass not just following the current.
Top comments (0)