When to Use AI in Game Development (And When Not To)
What's the first thing you did with ChatGPT? For me, I asked it to do a few deeply creative things - write me a lullaby based off the items on my desk or write me a murder mystery story. It wasn't great. But then I had it debug some code and write an email, and suddenly it was actually useful.
After over a year of using LLMs daily, I have a continuously growing appreciation for how good they are at finishing thoughts — but not starting them. This is exactly how I use them for gamedev: not for coming up with ideas or creating assets, but for helping me get from point A to point B when I already know both points clearly.
The Clarity Test
Your ability to clearly explain what you want to AI is the perfect signal for whether you should be using it at all. If you can precisely describe both your current state and desired outcome, you're probably working on something technical where AI can help. If you're struggling to explain it clearly, you're likely in creative territory where you need to do the work yourself.
This realization has made me a better game developer in two ways. First, it constantly challenges me to truly understand my game's mechanics - I can't effectively ask AI for help if I can't explain things clearly. Second, it's become my canary in the coal mine - when I'm having trouble describing what I want to AI, it's usually a sign I'm trying to outsource creative work that I should be doing myself.
When Not to Use AI: Creative Core
Yes, AI can make art. It can generate sound effects. It can even design game mechanics. But should it? Nah. Every time I've tried using AI for the creative parts of gamedev, it's been a waste of time. The output is noticeably soulless, and more importantly, it doesn't help me grow as a developer.
There are two big problems with relying on AI for creative work. First, if you don't come up with core ideas yourself, you won't care enough to see them through. Second, if you don't understand how something works because AI made it for you, you can't fix it when it breaks or effectively build on that knowledge in your next project. You're just accumulating technical and creative debt that you'll have to deal with later.
When to Use AI: Technical Tasks
What I consistently turn to AI for is the nitty-gritty technical stuff. Need help tracking down a weird bug? Want to optimize some slow code? Looking to refactor something messy? This is where AI becomes invaluable as a solodev.
Game balancing has been surprisingly great with AI too. Not for vague stuff like "make this more fun," but for specific tweaks. When I can tell it "here are my combat variables, help me adjust them to make this boss fight harder but still beatable," it gives useful suggestions and helps me spot connections between mechanics I might have missed.
Finding Your Balance
AI is going to keep getting better and that's great for gamedev. Just be honest with yourself in how you engage with it - are you using it to solve technical problems or trying to shortcut the creative process? Let it handle the technical heavy lifting - that's where it shines. But keep the soul of your game in your own hands. Make sure AI is inspired by you and not the other way around.