AI Will Not Democratize Game Development

I keep seeing these arguments that AI tools in some form are going to open up game development to the masses.

Newsflash: Anyone can make games already.

There are free engines, free online tutorials, large and thriving communities devoted to helping people get better at making games. There are platforms that vastly simplify game making and commercialization. There are all kinds of third party tools. There are free games with source code and libraries of assets that people can adapt and build from.

All it takes to make games today is time, commitment, a basic technical literacy, and access to the internet. The Cambrian explosion has already happened. Everyone can make games.

So, AI isn’t going to make game development any more available to people than it is today. If you have the time, commitment, literacy, and access to use AI tools, you already have all the ingredients for video game development. If you’re missing any of those things, AI is not going to help.

But, people will argue, AI will save people time! They can make bigger games faster, more polished games than they would have been able to afford, better graphics than they could have made on their own. Programming won’t be a barrier, anymore!

Okay, let’s assume for a moment that all of that is true, in spite of the obvious flaws and easy critiques. What you’re talking about is analogous to someone souping up a Honda Civic and then trying to compete in Le Mans or NASCAR.

The game industry is not some blue ocean. There are already dominant games in every genre. There are already existing platform holders, with hundreds of pages of regulations. The open platforms are already flooded with games. Every gaming community is currently being served by someone. You might have a fast car, but that doesn’t mean you can race.

For every hundred brave indies using AI tools, there’s a AAA team with professional coders and concept artists and QA. The things they want to use AI to do are already being done by larger, better funded teams, who also have marketing dollars.

But here’s the kicker. Games are feedback loops. Anyone can make a game, but making a good game is really freaking hard. It’s not the code; it’s not the assets; it’s not the scope of content. It’s getting the feedback right.

Game developers have evolved all kinds of tools and abstractions. We’ve made games in text and ascii; we prototype with cards and dice and hand drawn diagrams; we gray box and kitbash all the time, and the hardest part isn’t the art or how to write the code, it’s getting the fundamental interactions right.

And no AI is going to do that for you.

No amount of polished art is going to substitute. More bad content is worse, not better. And AI can’t QA. It doesn’t have judgment; it doesn’t have empathy; it can’t tell the difference between two different experiences as long as they are both “functional” within its rule set.

I’ve been saying for decades, we just need a program that can do game design. The only problem is how to design it…

So, this fantasy boils down, at some level, to the notion that people without the time or the commitment or the literacy or the access will somehow be effective when AI becomes available. And that just seems absurd to me, because the people who already do have all those things are making games right now.

There were over 10,000 new games on Steam in 2022. iOS and Android gaming stores already support more than 100K apps, each. There are more games being released on each of these platforms every year than any human being could play.

It’s literally impossible. There isn’t enough time.

This is what you’re competing with. If you don’t have the ability to make compelling games now, you’re not going to train an AI to make them. Even good games get lost in the vast deluge that is the game market today.

I'm not saying that AI won’t change game development; I’m sure it will. In fact, I know a fair number of very bright, successful people who are working on learning that problem space right now. Every hobbyist who was deterred before AI by how hard it was or how time consuming or how disappointing their first attempts were is still going to be competing with people who overcame all of those barriers.

So, no, AI is not going to democratize game development. Game development is already available to practically everyone (*first world privileges apply). Game development is still going to be hard. 90% of what gets made (if not more) is going to be crap. Even the best games are still going to struggle to be commercially successful.

Making fantasies that people can live in is only easy in your imagination. Making them in formats that people will play, much less pay for, is an entirely different order of problem, one that no AI is going to solve.