AI Use in Games

As I am nearing the point where I hope to offer the fully playable version of “Soundtrack of My Life” in the next few months (hopefully), I felt it was important to address the use of AI in my games. There are plenty of arguments in the development world about whether a game is truly developed by someone if it uses AI, whether it is legitimate, etc. So I will come out and say it: I do use AI in my games. There, I said it!

In fact, I use AI in many facets of my game design. I use Claude Code to help with the design tasks, writing code, connecting assets, etc. Then I use NovelAI for art generation and background generation. I am not ashamed of this. If you believe that this makes my games any less “real,” I will understand if you don’t want to play them. But before you turn your nose up and walk away, I feel like I should provide more context for my use and why I don’t believe it is a bad thing.

Why Do I Use It?

The simple answer is because I am the only developer in the company. I am the only employee in the company. My flagship game, SOML, alone has over 18,000 lines of code and over 185 image assets. That is A LOT of content to be included in a game, and game design isn’t even my main source of income (yet). I am fluent enough in Ren’Py and Python (the languages used in most of my games) to read the code I create and hand-type plenty of it. But when you think about it, the idea of one person who does not make a living as a programmer trying to create a game of this size by myself, using only free time, could mean my games take decades to be released. From inception to release, the technology it was built on may not even be in much use (ask the people who thought Visual Basic would last forever). It would be awfully hard to make a game in that time frame.

But let’s think about it from the perspective of a book author (as I consider many of my games to be books turned into games). If I am writing a book, I am fully capable of writing it myself (and I do NOT let AI write my stories; that’s 100% me). But what about the other stuff? A book requires editors who spend hours upon hours reviewing, reading, rereading, and correcting it to make it coherent. Then it has to be placed on the pages of the book (or converted to digital format for e-readers) with any sidebars, section headings, etc. But the work doesn’t end there; if the book will have any graphics, the author would have to either draw them themselves or contract with an illustrator to do that. While I strongly applaud anyone who can do all of that themselves, I cannot do it myself.

My Approach – Collaborate, Not Generate

I do not just sit at my AI console and say, “Okay AI, make a game that does this.” That would create a terrible game (most likely). Instead, I consider my AI to be a developer with years of experience in almost every known programming language. I think about it as though I were a manager leading a developer or a team of developers to create or maintain a software tool. Instead, I draft out the entire story skeleton (or at least the key things I want to create). This draft in and of itself generally includes several pages of written notes, likely some flowcharts that show my design ideas, and some graphic mockups of how I want a thing to look. Then I take all of this to an AI I have explicitly trained on how I think and what I want. I will then say. I have created an entire project plan (using Obsidian Vault with a few special add-ons) with sprints, goals, etc. ONLY after all of this is created, do I start the project.

So when I finally do start the project, I have done a lot of the pre-build work. I then will provide all of this to the AI and THEN say, “Okay, AI, let’s start this concept using all of these notes. I want to write this in RenPy; we will use the following Python code structures…” etc. Even then, I do not just type a prompt and tell the AI to do stuff. I actively look at the code it produces (there isn’t a single page of the game code I haven’t read through and iterated on). My approach is collaborative and not sloppy. All I am doing is using a tool as a force multiplier.

If you have just read this and still do not agree with my use of AI, that’s okay. But I want to offer another example that might be easier to follow. EVERYONE has used online sources to generate code. Does this mean it is “ripped off”? Not really. Because this code was not the only thing written in the project. If someone had copied the entirety of an online code snippet from Stack Overflow, Reddit, or wherever, just slapped their name on it, and said, “This is mine!” then I would agree that they are ripping that person off. But that is not what often happens (especially not when I write code). Code is a language. A language is a series of symbols and lines that are put together to make a thought coherent, and so is computer code. A sentence like “The light appeared in the forest at midnight.” may appear in hundreds or thousands of books, stories, and movie scripts. But does that mean that the person who wrote that was plagiarising someone else? Instead, it describes a thought/image that a person can express and other people can understand and interact with. Code with AI uses known symbols and lines to create a coherent idea.

The use of AI is not much different than what I described. I am using AI to take my ideas, thoughts, and plans, and turn them into something I can share with others. I could eventually end up in the same place at the end; if I do it all myself, it will just take SO MUCH LONGER. My AI usage is much more akin to a person working with a team to deliver something more quickly than they could alone. But instead of just assuming the code is correct, I will cross-check it and painstakingly test ideas. I also don’t let the AI write the story for me; instead, I depend on it to make the code design that would take me much longer to do by hand, and then I add my creative elements on top of it.

Isn’t AI Dangerous?

Well, not exactly. I often explain it like this. Every tool is only as useful or dangerous as the hand that holds it. Don’t believe me? Then go pick up a hammer. The hammer’s creator designed it to let a person hammer nails into boards, remove stray nails, pull things apart, etc. If you were to go out into the world, you are likely to find millions of hammers. They would be scattered around job sites, stored in toolboxes, hanging in garages, or rusting in a dumpster. An overwhelming number of hammer users will use hammers for the exact purpose for which they were designed.

But what if you take a hammer (maybe you borrowed it), try to nail a board in, and end up busting your thumb? Is it the hammer’s fault that you hurt yourself with it? No, the hammer lacks the sentience to judge whether you intended to hit your thumb. You could say it is because you were not familiar with how to use a hammer. Or maybe you know how to use a hammer but overextended your hand. Maybe a dog barked and distracted you. In all these cases, the hammer was only the tool.

Even darker, it is highly unlikely that the hammer’s original creator built it to bust someone’s skull in. If a person with a sick mind decides to take a hammer and go on a killing spree with it, they are not using the tool for its intended purpose. You cannot blame the hammer for being used by a person with a depraved mind to do unspeakable things.

AI is a hammer in its own right. It was created to solve problems (such as a hammer being designed to pound in nails). Anyone with access to AI can use it for whatever purpose they choose (within the AI’s AUP). If a hammer had an AUP, one of the things it would likely say is “Don’t use the hammer to bust your thumb” or “Use this on nails, not your neighbor’s face.” Even still, a bad actor can completely ignore the AUP and say, “Too bad, I am going to use this to destroy my hand,” and there is only so much the hammer’s creator can do in that situation.

Ergo, AI is in and of itself NOT a dangerous thing. It is only dangerous when it is used improperly. I am using it to make things faster and more robust than I could otherwise. I hope you understand this mentality. Perhaps, if one of my games becomes an amazing success (hey, you reading this, help me with that!) I can hire a whole fleet of programmers and illustrators. But until that day, I will use the tools I have to build something amazing, and I hope you will enjoy playing it as much as I enjoyed writing it.