I started learning to code in 2017 and have always wanted to build games. Before relying heavily on AI, I enjoyed taking an idea and solving it step by step—for example, figuring out how to represent inventory items, handle dragging, detect slots, and gradually turn those smaller problems into a working system.
Since around 2022, I've started using AI for much of my coding. Instead of feeling like I'm improving, I often feel like I'm getting worse. I'll ask it to build something, run into problems, ask it to fix them, and eventually lose interest. It feels less like using a tool and more like entering a cheat code, because the part I enjoyed was designing and discovering my own solutions.
Someone I know runs a successful company that builds websites and apps largely with AI, and they keep telling me that AI is simply a tool I need to adapt to. I understand that perspective, but I'm struggling with motivation and creativity. If an inventory system, navigation system, or other feature can be generated from one prompt, it's harder for me to feel excited about building it.
I'm a third-year game development student and haven't worked in a professional software company yet. For people who learned to code before AI became mainstream, what keeps you motivated now? How have you incorporated AI without losing the problem-solving and creative parts of programming?
5 Answers
The important distinction is whether AI is replacing your thinking or extending it. Instead of asking it to build an entire feature, start by describing the goal and discussing the design: what data structures make sense, what edge cases exist, and how the systems should interact. Then implement and review the solution in small pieces.
That preserves the part of programming where you form a mental model and solve problems. AI can help with research, boilerplate, and alternative approaches, but you should still be able to explain and evaluate everything it produces. Treating it like a junior developer or brainstorming partner is very different from blindly accepting its output.
AI isn’t automatically cheating, but using it without understanding the result can weaken your skills. In professional work, the valuable part is often deciding what needs to be built, understanding the constraints, and maintaining the system afterward. Code generation can speed up implementation, but it doesn’t remove the need for design, testing, debugging, or a mental model of the software.
A useful compromise is to reserve AI for boilerplate, documentation searches, repetitive refactors, and exploring unfamiliar APIs. Write the interesting algorithms and core game systems yourself, or have AI provide suggestions without directly generating the final implementation. That way you gain speed without giving away the parts you actually enjoy.
Some developers genuinely enjoy the architecture and outcome more than manually typing every line, while others get satisfaction from the craft itself. Neither reaction is wrong. AI may make work faster, but it can also remove the human conversations and discovery that made programming rewarding.
For a game development student, it may help to create projects with explicit constraints: no AI for the core mechanic, limited AI use for research, or handwritten implementations of selected systems. Work with classmates, discuss designs, and show unfinished experiments to other people. Creativity usually grows from making many imperfect things, not from waiting for one perfect idea or asking a tool to provide the entire project.
There have been several versions of the programming profession, and each one moved further away from the simple, low-level work many people originally loved. Modern development already involves choosing libraries, frameworks, and services rather than understanding every layer from the bottom up. AI pushes that abstraction even further.
That doesn’t mean the underlying skills are worthless. Algorithms, architecture, debugging, and understanding how computers work still matter. If you miss that feeling, make room for personal projects where you deliberately write the important parts yourself—perhaps a small engine, simulation, renderer, or game mechanic. Your job and your personal craft don’t have to use the same workflow.
You also don’t have to adopt the most aggressive AI workflow just because companies are promoting it. Use it where it genuinely helps and skip it where it interrupts your flow. Many experienced developers use it for boilerplate or large mechanical changes, then spend significant time reading, testing, and correcting the result.
The main danger is outsourcing judgment. If you can still explain why the system is designed a certain way, identify when the generated code is wrong, and maintain it later, you’re using assistance rather than abandoning the craft.
That describes exactly what I’ve been doing wrong. I used to break an inventory system into questions and solve them one at a time. Recently I’ve just asked for the whole feature, then become frustrated when the result fails. I think I need to use AI to discuss and refine the idea while keeping the overall design in my own head.