Like many people, I use AI as a tool to support me, and it’s an incredible resource if you want to learn how to code. It makes the programming world much more accessible, even for people with no prior experience. When you’re stuck or can’t find your mistake, AI can help you move forward. No more hours of endless searching on the internet just to find the right answer.
But there’s also a trap. The temptation to let AI write the code for you is very strong. And that’s the point where learning stops. I found that you no longer learn from your mistakes, you don’t build an understanding of the constructs, and over time you lose sight of what each part of the code really means. With small or simple examples this might still be manageable, but as soon as projects grow longer or the structures become more complex, your understanding starts to fade, leaving you back where you once began.
The real value of AI, at least in my experience, lies in using it as a mentor rather than as a substitute. Asking why something works, or letting it explain concepts in different ways, often gives you insights that a traditional tutorial cannot. When you take the time to debug with AI instead of copy-pasting a solution, you still gain the essential problem-solving skills. It becomes less about shortcuts and more about building a deeper foundation.
So where exactly is the turning point? At what moment does AI stop being a helpful assistant and instead become a barrier you need to overcome? Perhaps the answer depends on your own discipline: the moment you stop asking questions and start outsourcing your thinking, AI is no longer your ally but your crutch.
Thanks for your contribution! I think LLMs should be really used for coding tasks as many coding is just repetitive work. However, you should first understand this task. Only start to use a LLM for coding when you already have done this coding for yourself and fully understand the AI output.