I have an idea for a Python project and want to turn it into something I can keep developing over the long term. My Python experience is limited, and I'm also still learning Git, deployment, and the rest of the development workflow.
I originally used an AI coding assistant to build the project quickly, but after a few weeks I realized I no longer understood much of the code or felt in control of the project. Although it produced a working program that I could deploy, the experience felt discouraging because I want to understand what I write and continue coding as a career—not just launch this one project.
I've started over with a clean repository. Should I avoid AI-generated code entirely and learn by implementing everything myself, using documentation or occasional questions only when I'm stuck? Or is there a productive way to use AI while still developing real programming skills?
4 Answers
Use the AI as a mentor instead of an automatic code generator. Explain your design, reasoning, and tradeoffs, then ask it to challenge your assumptions or give you hints without immediately showing the solution. You can also use it like a demanding rubber duck: make yourself explain each decision clearly, and only request an example after you’ve made a serious attempt yourself.
You probably don’t need to ban AI completely. A useful rule is that you shouldn’t keep code you can’t explain. Try each problem yourself first, make small changes, read every line, and use AI for focused explanations or feedback rather than large blocks of implementation. If the assistant writes something, rewrite it or verify that you could reproduce it yourself. Your understanding should never fall far behind the code in your repository.
Since you want programming to become a long-term career, prioritize learning the fundamentals yourself. Start by attempting each feature, then consult documentation or ask specific questions when you’re stuck. AI can review your approach, point out gaps, or explain an unfamiliar concept, but letting it generate entire features makes it much harder to understand how the system works.
If your goal is genuine expertise, you need to spend time struggling with problems, testing your assumptions, and correcting your own mistakes. That discomfort is part of how programming judgment develops. An assistant that removes all of the difficulty can help you produce results while preventing you from building the skills behind those results.
That balance is probably where I’ll end up. I’d like to reach the point where I can implement things comfortably on my own before relying on AI to produce code for me.