I used to be completely obsessed with programming. About five years ago, I would stay up late experimenting in an unfamiliar IDE, even when I barely understood anything beyond variables and functions. I once spent hours chasing a C# error from a tutorial, only to discover that a capital letter needed to be lowercase. The frustration was intense, but solving those problems felt incredibly rewarding.
Recently, I became interested in programming again, but the surrounding culture feels different. So much of the discussion now focuses on AI replacing programmers or people deliberately avoiding AI to prove they are doing things the "hardcore" way. Neither side really appeals to me. If I want to build something, I can either relearn Python and work through every problem manually, or ask an AI for an answer immediately. Struggling purely for the sake of avoiding a tool feels artificial, but removing all the struggle also removes much of the fun.
I program casually and for enjoyment rather than as a career. Has anyone found a satisfying middle ground where AI can be used efficiently without taking away the discovery, debugging, and sense of accomplishment that make programming fun?
5 Answers
Working through a book or a well-designed course can provide the structure you are missing. After each section, build a small project from scratch and let yourself struggle with it before asking for help. There is nothing wrong with using AI after you have made a real attempt, especially if you use it to explain an error or critique your approach instead of generating the whole project. The goal is not maximum suffering; it is understanding enough that the solution feels like yours.
The difficult part of programming does not disappear with AI; it just moves. Generated code can contain subtle mistakes, poor assumptions, or advice based on unreliable examples, so understanding and troubleshooting still matter. You may find that satisfying difficulty in designing the project, making tradeoffs, and tracking down unexpected behavior rather than spending hours fixing a missing capital letter.
AI does not have to remove the fun. Treat it more like a faster version of searching through documentation than an automatic code generator. You can ask about a particular algorithm, request examples, or have it point out possible issues in code you already wrote. I prefer writing the implementation myself and using AI as a second set of eyes. That preserves the learning while avoiding hours of searching for a tiny detail.
Try stepping away from programming-content videos and spend more time actually building things. Conference talks, books, and structured courses tend to be more thoughtful than short-form programming content. Pick a language or project that genuinely interests you, work through the fundamentals, and then make something without constantly looking for the next tutorial. Use documentation, search, or AI when you hit a specific obstacle, but do not let them replace the process of exploring.
You probably cannot recreate exactly how programming felt years ago, because you are not the same person anymore. Instead of chasing the old aesthetic, figure out what parts of programming still appeal to you now. You can also choose a niche with unusual requirements or poor documentation, where you will naturally need to investigate, experiment, and debug manually. The challenge will come from the project itself rather than from artificially refusing every modern tool.
A useful compromise is to compare generated code carefully, line by line, and ask why each function or design choice was used. That turns the result into something you study instead of something you blindly paste.