Is self-studying coding with ChatGPT inefficient?

True, it's insanely fast lol but that speed is a trap

I wasted six months doing that too. The code GPT wrote for me would run, but since I didn’t know why it worked, the moment an error hit, I’d freeze up and just sit there helpless. Now I’ve changed my approach: I learn the concepts from books or official docs, and I only use GPT to ask things like “Why is this written this way?” or “What’s another way to do it?” It feels less like a code generator and more like a tutor.

Got a source? Or is that just a hunch?

It's normal that you can't explain the error at first. But you can just ask GPT again, 'Why is this error happening?' The problem is moving on without asking.

Oh, I didn't know that.

Well, that's a bit... Self-studying by asking GPT isn't inherently inefficient, but what you're doing now seems closer to "copy-pasting" than "understanding." When an error message pops up, before asking GPT for the answer, try reading the logs and stack trace line by line. That one habit will make the difference in your skill a year from now. Seriously.

A mix of 30% lectures and 70% GPT worked best for me. Lectures were boring and I couldn't make any progress, but GPT's instant answers kept a good pace.

I disagree. Why are lectures inefficient? There are far more cases where one well-made lecture is better than asking GPT a dozen or so times. Isn’t it just that you don’t want to watch lectures?

When you get code from GPT and ask it to "explain this line by line with comments," you end up understanding it too. Surprisingly many people don't know this.

I feel the same way. These days, I don’t just use GPT’s answers as-is; I’ve made it a rule to struggle on my own for 30 minutes first, and then ask. The time I spent struggling doesn’t feel wasted—if anything, it sticks with me more.