The creator in AI game creator is you

Deciding what the game is. Deciding what it looks like. Building the world. Making things behave. Playing it, noticing it is not right, and changing it. Doing that last step enough times.

Only one of those five is code. It happens to be the one that stops most people, which is why an AI that writes it is genuinely useful. It is not the one that makes the game good.

A tool that claims to do all five is claiming to have taste about a game it has never played. The reason so many AI-generated games feel the same is that they were all created by the same taste, which is to say none.

Director 1 writes GDScript in your project. Movement, inventory, dialogue, enemies, objectives, saving, menus. You describe what should happen; the code appears where you can see it.

You build the world by hand in the builder. Terrain, water, weather, buildings, characters, light. Placed, not described.

And you play it. Every time. That is the part that cannot be delegated and it is the part that determines whether the thing is worth playing. Sending work back because a section drags is the normal loop rather than a sign of trouble.

The assumption baked into a lot of tools for non-programmers is that the person also needs protecting from decisions. Simplified options, a template to pick, an outcome chosen for them.

That gets it backwards. Someone who cannot write a state machine can still tell you the enemy is too aggressive, the level is too dark, the middle third is boring and the ending arrives too fast. Those are the hard calls and they are made by playing.

So the app asks a lot of you and very little of your technical background. There is no template list to pick from, because a fixed list of game shapes is the fastest way to produce a game that was already made.

Desktop app, Mac and Windows, downloaded and installed. Godot underneath, open source. Your project is a real Godot project on your disk.

The game runs locally on the bundled Godot. Building needs an internet connection since the AI and world data come over the network; playing what you built does not.

Free to download with a free plan. Indie is $10 a month, Pro is $50, and both buy usage rather than a different engine.

Finishing a game is mostly a pause menu, key rebinding, a settings screen, a save that works, and cutting the section you spent three weeks on because it drags.

Every one of those except the cutting is code, and code is the part written for you here. That genuinely changes the odds, because this is the phase where most projects quietly stop.

The cutting is still yours and it is still the hardest thing. No tool will tell you that the level you are proud of is the reason people stop playing. Only playing it, and watching someone else play it, will do that.

The first game teaches you the process. The second is where the tool starts paying off, because you already know what you want and you are no longer learning the loop at the same time.

That is worth knowing at the start, because it changes what a first project is for. It is not meant to be the game you have been imagining for years. It is meant to be finished.

Small, complete, and played by someone else. Then the big one, with everything you learned about your own taste, which is the only thing that actually improves.