Back to home

Games, news and stories

Machine translated · For reference only

Don't worry—AI games pose no threat to indie game developers.

Russell

page_bg_raw.jpg.png

Recently, AI has become increasingly capable of creating games—it can handle coding, art, storytelling, and even produce a playable prototype. I’ve also created two or three AI-generated games and published them on CrazyGame, even earning some ad revenue in the process. As a result, many indie developers are starting to worry: If others can make a game in just a few days, while we spend years on one, will anyone still play ours by then?

As an indie game entrepreneur—and as a game programmer who has worked on multiple projects—I’m actually not that worried.

It’s not that AI won’t affect us; rather, I think that for independent developers who truly want to create their own work, the fact that AI is getting more and more powerful isn’t necessarily a bad thing. We don’t need to compete with it over who can produce the most content in a day, and we certainly don’t need to feel that our work has no value just because others can do it faster.

I’d like to share my thoughts on this from a few different perspectives.

AI art: Looking beautiful and being comfortable to look at for a long time are two different things.

At this stage, I think the art in AI games is still a major weakness.

Of course, AI can generate beautiful images—there’s no arguing that. But a game isn’t just a single image, and players don’t just glance at it for two seconds before swiping away. Characters, environments, interfaces, and animations all need to come together to keep players engaged for a long time.

Some AI-generated images look quite polished on their own, but once they’re incorporated into a game, they start to feel out of place. There’s an overload of detail—everywhere is crammed full—and the characters and environments seem to exist in separate worlds. They might look good at first glance, but the longer you play, the more jarring they become.

I know some people will say that as long as a game is fun, that’s all that matters. But players have to look at the visuals the whole time—if the art makes them uncomfortable, why should they have to put up with it just to keep playing? There’s no shortage of games on the market.

Players might even think, “I like this gameplay, but is there a game with similar mechanics but different art style?” In that case, this AI-driven game actually serves as inspiration for others’ work and proves that while some people might enjoy this gameplay, the game itself may not necessarily retain players.

Creating a single attractive image is not the same as producing a set of game art that people are willing to look at for hours.

As for having AI create art that doesn’t look obviously AI-generated, I still think it’s difficult—or at least, the cost isn’t as low as one might imagine. It requires repeated screening, revisions, and stylistic consistency, as well as setting up complex workflows—and in the end, the time spent may be comparable to that of manual production. If someone can solve these problems effectively, it’s because of their skill and judgment, not just a single prompt.

So, at least for now, I don’t think AI has easily solved the problem of fine art.

But I don’t want to console myself by saying, “AI isn’t ready yet.” It will keep getting better, and the shortcomings we see today may very well be resolved in the future.

So let's assume that, in the future, AI will be able to handle every aspect of a game with such precision that players won't be able to tell it was created by AI. I still don't think we need to worry too much.

I often tell my friends:If you don't like Ubisoft's "canned" games today, chances are you won't like "canned" AI games in the future either.

Ubisoft excels at creating games with beautiful graphics, rich content, and well-rounded systems. At first glance, there aren’t any major flaws, but with the same familiar missions, maps, and progression systems repeated game after game, it can get tiresome after a while. Of course, I’m not saying every Ubisoft game is like this; I’m referring to that experience where “it has everything, but nothing feels fresh.”

Personally, I think a lot of AI games will face this problem in the future. In fact, I’d go so far as to say that if developers focus solely on efficiency, maturity, and safety, then in my view, the ceiling for AI games will be this kind of “Ubisoft-style” cookie-cutter experience.

Because ideas that go against the consensus are rare to begin with. AI might be able to propose ideas that go against the consensus, but even if it does, developers may not necessarily agree—this requires the developers’ own judgment.

For example, if a proposal aligns with our experience of successful games, we tend to find it reasonable. If another proposal seems strange—something we’ve never seen before and don’t know if anyone would like—we might ask the AI to tweak it further. By the time the revisions are complete, it ends up being something everyone is familiar with again.

So the problem here isn’t necessarily that AI lacks creativity; it could also be that the people using it are reluctant to accept unfamiliar ideas. In the end, everyone chose the option that seemed correct, and naturally, the games began to look more and more alike.

What makes indie games so interesting is that there are always people willing to take on projects that seem a bit off the beaten path—ones that haven’t yet been proven to succeed.

Not every indie game needs to pioneer a new genre, but if someone is making an indie game just to quickly cobble together an existing one, then I think they really should be worried about AI. Because when it comes to that, AI will likely do it faster than they can.

If you have an experience you want to create—an idea that others might not agree with but that you’re willing to try—then I actually think AI can truly help you, for example, by helping you overcome potential technical challenges or identify flaws in your design.

Now that technology has leveled the playing field, we should focus on conducting more experiments rather than producing more junk.

As a game programmer, there’s one thing I have to admit: the barrier built solely on the premise that “I can write code, so I can implement this feature” is disappearing.

I know many programmers still believe that AI can’t replace them. I don’t really want to argue about this—let them think what they want. What concerns me more is that more and more tasks that used to require specialized technical skills to get started can now be attempted with the help of AI.

Of course, there’s a difference between building a prototype and delivering a stable game, and complex engineering isn’t something that can be accomplished with just a few words. But for a small team like ours, AI has made it possible to experiment with ideas that we previously couldn’t afford to try.

In the past, when I had an idea, I had to learn the technical skills, do research, and then spend a long time bringing it to life—and I might run out of energy halfway through. Now I can put together a version much faster and see right away if it’s actually interesting. I think that’s a good thing for creators.

The question is, what should we do with the time we save?

Should we let AI quickly replicate a bunch of proven formulas, plug in some content, and sell them? Or should we use it to test ideas that we couldn’t verify in the past? I’d much rather do the latter.

Now that we’ve achieved technological equality, we should conduct more experiments rather than rush to produce a pile of junk.

Try an interactive approach you’ve never seen before, create a mechanism whose viability is uncertain, or explore a new technical solution. These efforts may fail, but even if they do, you’ll learn why that particular direction didn’t work—which is far more meaningful than constantly churning out reskinned content.

The ideas here don’t have to be particularly grand. A small mechanic, a unique gameplay rhythm, or a new way for characters to interact are all worth giving a serious try. Innovation isn’t just about creating “the world’s first [type of game].”

I think the true future of indie games lies in more creative people finally having the ability to bring their ideas to life, rather than everyone using the same set of tools to produce the same kind of products.

Keep good taste, and don't be stingy with tokens.

So, of course, I’m not against using AI; I just hate the junk it churns out in bulk. That’s a waste of this new revolution—what we should be doing is making good use of it.

AI knows everything in the world; we can design things to discuss with it, write code to have it implement, and use it to explore art and storytelling. But once these things are created, it’s up to us to decide whether we want them—we can’t just accept whatever it gives us.

At times like this, having good taste is very important.

By “taste,” I don’t just mean knowing which image looks good—it also means knowing what kind of experience you want your game to deliver to players. If a design proposal looks complete but doesn’t fit the game, it shouldn’t be kept. A feature might be easy to implement, but if it’s unnecessary and there’s no reason to force it in, it shouldn’t be included. It really comes down to the fundamentals of game design—the knowledge we’ve gained from books and hands-on experience is even more important now. I believe this is the “quality” mentioned in *Zen and the Art of Motorcycle Maintenance*—something AI simply doesn’t possess.

Otherwise, AI can easily lead us astray. It presents a system, and if we think it’s decent, we incorporate it; it writes a long block of text, and if we find it rich in detail, we use it; it generates a beautiful image, and even if it doesn’t fit with the rest of the material, we can’t bring ourselves to discard it.

In the end, the game kept getting bigger and bigger, but I couldn't really explain why those elements were necessary.

Good taste, first and foremost, means not letting AI lead us astray.

We use AI to improve efficiency, not to let it dictate the direction of our work. We need to know for ourselves what to keep, what to remove, and where improvements are needed. Only by using it this way can we ensure our work doesn’t end up looking completely different as we create it.

One more thing: I don’t think we should be too stingy with tokens. Whether it’s a game or any other product, if you want to create something great in the future, you’ll definitely need a lot of tokens to refine it,

Good taste serves another purpose: it can help us recognize an idea that others might find strange but that we feel is worth trying. AI can generate many options, but determining which ones are interesting and which are worth pursuing still requires judgment—and actually implementing them in the game to see how they play out.

Sometimes, a new way to play or a new product category can start right here. Not because AI has told us it’s bound to succeed, but because we’re willing to give it a try and keep refining it.

Let go of your anxiety, maintain your good taste, and focus on making games

All in all, I don’t think independent developers need to spend all day worrying about what AI might come up with next or which technology has “crashed” again.

It’s true that AI will transform industries and bring more competition. But we can also use those same tools—and apply them to the things we really want to do.

If the competition is just about who can write code faster or cram in more content, it might get harder and harder. But I don’t make indie games to prove that I’ve written a few more lines of code than an AI. I still hope to create a game that I enjoy and that others will want to play.

Have your own ideas, maintain good taste, and use these tools to create a great game. I think that’s a better use of your time than worrying about how many games others have made in a day.

Let go of your anxiety, maintain your good taste, and focus on making games.

Let’s encourage each other.

Sign in to save

评论

0 / 5000 个字符

    暂无评论