My latest game BossBattle[1] (html5, have a play) uses AI for graphics, some strobe effects, a C64 loading screen shader.
I have decided to lean in to it and I will document all the places I use AI in the game on my blog[2]. Not everything works, notably 3D assets[3] and sound effects.
There is a lot of human content… i paid for a lot out of my own pocket and have limited budget. It started in 2021 before chatgpt. LLMs cannot do everything and that’s not the purpose.
Generative AI makes me as an solo indie dev able to make the game. Without the AI the game wouldn’t exist
I spent quite a bit of time making games/social WebXR stuff in Three.js. There is unfortunately a ceiling to what commercial success you can have there. Sure there's all the .io games which see a bunch of kids playing games from their school computers, but as much as I'd like to see it, web games aren't going to be as impressive as native PC/Console games. It's not really technology, mostly just the market isn't there.
The models are pretty decent at building simple Three.js/Phaser games, but if you want to work with Unity/Unreal/Godot you're going to need a MCP or other tool to get them to work with the engine's tooling/context. I just so happen to work on one for Unity https://bezi.com
I will say, while I think the current models are very impressive with generating code for most game mechanics. They are still terrible at spatial awareness. Gemini Pro 3.1 is showing some promise here, the latest Opus/Sonnet models are...ok. But there's still a lot left to be desired. You also still really need to know how to make games both creatively and technically to pull off prompting a game into existence.
So are you going to vibecode your way to the next SimCity / Civ without knowing some game dev? Probably not right now and I think that's for the best. People want games that are creative and unique. But a passionate hobbyist who has never made a game, knows some programming, and has a vision for a great game now has an amazing tool set to build their dream game and that's pretty cool!
Shameless plug - and nothing so grandiose as SimCity but I built a pretty substantial 2D/3D blindfold trainer chess game. It's by no means "vibe coded" though, and there's a fair bit of manual work around the 3d modeling that I had to roll myself.
Even with that I'd still say 70% of the code was written using LLMs with the opencode agent.
I wanted to create a game (Pre-AI) and thought it would be easy. Turns out it is not at all.
Sure you might get lucky with the next Angry bird, but there is a whole range of skills required to make a good game, and actually get people to play it.
I even read a couple of books which gave me a better understanding of how out of my depth I was. Books were "Theory of Fun", "Achievement unlocked" and "The art of game design".
AI here appears to be accelerating the ability to see those gaps faster. I think without the understanding of those gaps anything created is going to be lacking.
Call it whatever you want, the game wouldn't exist without them doing something about it. And because they did something about achieving their vision other people get to play the game.
How about we just call these people the game Producers instead. Thats what a producer does anyway right? They make decisions of how the game is built, what goes into it, ect, ect, ect.
Oh please, just stop being silly. Being a game dev means you're building games which means you're thinking about the mechanics of the game and what makes something fun to play. The programming is not "game dev" and offloading it to an AI agent doesn't make someone any less of a game dev.
Totally! I'm sick of not being able to see what tools they use. In fact, I feel that we should also start tagging devs that use an IDE instead of properly coding games in notepad.
If you want to see what the HN game dev community is building, I’ve been curating the list here (https://hnarcade.com) for the past months.
The output is quite impressive. And having spoken to a number of the developers, it does seem like AI has had a massive impact on delivering their ideas.
I think the article is being s bit disingenuous. The real problem is triple A developers and publishers pushing for it to replace a not insignificant amount of creative work.
While still over charging everyone and scalpling every $ from everyone with micro transactions and game mechanics that need xp boosters.
>Or is the underlying 'brainrot' actually destructive to creative potential?
Creativity comes from constraints. Writing code hasn't been the hard part since the 90s. Deciding which things make for a good game and are worth spending your limited time on is where fun comes from.
AI makes it easy to spit things out, but it doesn't make things fun or good at all.
Time will tell! Certainly AI is unlocking a lot of stuff, a huge number of people who did not have the time or the special skills before to create games will do so. I can think back to my own start in making computer games, using Amos on the Amiga (vastly easier than C or assembly).
It's a separate question whether anything actually good will come out of it. It's incredibly unfair to look at any particular project and say: what, another clone of this or that done idea? Very few things are original in any time. Certainly I didn't make anything particularly original all those years ago. But, soon, there should be really something, if there's really something there. If it's not just burning tokens to copy older ideas. And we'll know it when we see it, this amazing thing that would not have existed otherwise.
I wish that people would just stop with the "vibe coding" thing.
If you want to get quality results from an LLM use a quality frontier model (I recommend Opus 4.5 thinking) in an agentic Plan -> Agent -> Debug loop inside of Cursor. Roughly 90% of the hate that gets assigned to AI anything is a direct result of the absurd notion that taking the human completely out of the loop is a valuable goal. In reality, it's expensive and almost guaranteed to produce crap.
If you treat LLMs as pair programmers and split your implementation into a set of sequential tasks of a reasonable scope, you can use Unity or Unreal or any number of JS engines built on ThreeJS to produce things that are worth playing.
I would strongly argue that pairing with Opus to write your controller code while you take primary responsibility for interacting with the UI sounds exactly like how you should proceed if you care about the end result.
Im using UE5 which is arguably GUI heavier than Unity.
You can get around a lot of the GUI-heavy stuff by using C++ in preference to Blueprint, and/or developing some tools to help you decompile/recompile Blueprints.
It has a pretty cool remote control plugin you can install which can be used to simplify a lot of test cases through automation.
I have a relatively large amount of experience with UE4/UE5 and C++ though, so it's probably not for the absolute beginner or the faint of heart.
I've build 10+ games which I am ashamed to show :)
Straight JS/html/css front-end with zero dependencies works well.
Ask for a node.js backend and can be instantly deployed as client/server or straight to html - multiplayer feels trivial.
C# Monogame works well for something heavier.
You can actually edit Unity scenes directly using the LLM as they're a readable text file which works ok, but Unity is bloatware when you can code it all yourself (it's an absolute nightmare of inexplicable bugs, do not use it. After updating to 0.62f from 0.48f my clang compiler now segfaults while building Webgl - luckily my team mate can do the builds)
The key is building exactly and only what you want and need. Make your design lean, suit the game as you are actually building it not a theoretical overengineered masterpiece - refactors are cheap later, but bloat will kill your project.
I've had my own rollercoaster relationship with Unity over the past decade, but telling people to roll their own game engine so that they can finally make a game is almost universally terrible advice.
People who want to build game engines should build game engines; people who want to build games should absolutely use Unity, Unreal or Godot in no particular order.
It's no different than needing to build a web framework so that you can make a website. The people who do it are often not even aware that they are procrastinating.
Big plus for html/css/js, mostly Pixi 8 or around (I also have a couple threejs). Vanilla JS. I did this by hand before, but having the LLM tweak around the code and styling while I handle more gameplay related things makes this doable (otherwise I just would not have enough free time)
> Nobody's sharing their work, because they used AI.
Really. What a stupid statement. As if AI bros aren't screaming all over the internet telling people their vibe-coded projects.
I do use AI for code, but I really don't know why AI users seem to have some sort of victim mentality when there are literally billions of capital stacking on their side.
I strongly disagree. Let me try and change your mind.
Let's say that you have been building your passion project; it could be a game, it could be a synth, it could be a mobile app. You were a year into development already when LLMs blew up, and you very quickly realized that you could leverage them to move much, much faster. This allowed you to tackle a long list of features that realistically would have never left the daydreaming to-do list because in reality, it'd have been a minor miracle to just get to v1.0 much less do all of the cool stuff you thought of along the way. This is the real power of the agentic workflow: in experienced hands, it's an incredible force multiplier.
And yet, even while your churning out features and knocking out bugs that you never would have found on your own without users being upset, the world gets progressively more and more pissy about this moment's Scarlet Letter. Suddenly mentioning that actually, you're getting amazing results from LLMs makes a significant percentage of people irrationally angry, leaving comments that would otherwise be appropriate to shame deviants and cheaters.
Suddenly, you're faced with a really awful realization: you'd love to loudly and proudly show the world how your LLM-assisted dev workflow has allowed you to produce something you honestly might not have ever finished in 2023, but if you do then people are going to write off what you've done as slop before even looking at it, and since you were actually thinking that maybe what you were building might have some legitimate commercial viability, you can't actually advocate for sanity and transparency without taking a serious hit to your product's success. By extension, this means that it might cause real damage to you and your family.
What I am describing is a chilling effect, and it is real. All of the billions in VC have literally nothing to offer someone who just wants to build cool shit and can suddenly do so faster and often better.
At the very least, consider that there is a tangible cost to telling the world that the thing you created exists at least in part because of AI but there is no such thing as people being more likely to pay for something simply because AI was involved. There is only downside. That is the chill.
It's not about victimhood. It is about not fighting battles that can only lose you ground.
i don’t like ai art bc it lacks and sort of “soul” or “human touch” (nebulous and subjective but iykyk) games are just another form of art, so why would i waste my time with just another form of slop?
Similar. I want to see games made by humans who have put in the effort and taken the time to build something good. I don’t want to see the market flooded with low effort AI slop.
You should consider that there is a rather large gulf between low-effort crap and using agentic LLMs to make more sophisticated games faster before you downvote me.
It's just not black and white, and to treat it as such devalues the conversation.
I mean all the major hurdles of making a game aren't really helped much by AI...
AI has no training data on complex logic and systems so you gotta do that all yourself.
It definitely doesn't get anything visual right really.
There isn't large amounts of automated testing you can setup ahead of time for a lot of game-play so the AI can't iterate on it to make something work it'll just be hopeless.
The art is also going to all be really derivative plagiarism overly averaged scammy looking stuff. So that's basically an insurmountable hurdle. No unique style.
>I mean all the major hurdles of making a game aren't really helped much by AI...
I have been using Claude Code to develop a game in unreal engine. It is fricking amazing. Its like hiring someone with 10 years experience to work for you. I am really impressed by how it know game patterns.
>It definitely doesn't get anything visual right really.
Sometimes it struggles to things things right visually, other times it nails it!
I have been using an MCP from gemini image 1.5 to generate my icons. And once it go my styles down, after 20 experiments, it does really good. Notice: It uses high quality by default which will burn up your credits. But if you turn down the image quality to low, it cost around 3 cents an icon.
>There isn't large amounts of automated testing.
Some things can be easily automated for testing. But other things require play testing.
>The art is also going to all be really derivative plagiarism
I am just using it to generate icons, and it does great. For 3D artwork I either use things from the FAB Store, or I pay a team of artists in Pakistan to do it.
Overall I say it is the equivalent to have a senior dev on your team, for 100 bucks a month
My latest game BossBattle[1] (html5, have a play) uses AI for graphics, some strobe effects, a C64 loading screen shader.
I have decided to lean in to it and I will document all the places I use AI in the game on my blog[2]. Not everything works, notably 3D assets[3] and sound effects.
There is a lot of human content… i paid for a lot out of my own pocket and have limited budget. It started in 2021 before chatgpt. LLMs cannot do everything and that’s not the purpose.
Generative AI makes me as an solo indie dev able to make the game. Without the AI the game wouldn’t exist
[1] http://epicwin.team/play/solo/BossBattle/ - (public beta) .
[2] https://generative-ai.review .
[3] https://generative-ai.review/2025/08/3d-assets-made-by-genai...
I am excited about game dev with AI, but the games you posted are kind of a joke.
My kids made similar games with Claude code in js.
Was hoping to see some serious indie games, but these looked pretty terri-bad.
Is anyone building the next SimCity, Civilization, etc.?
I spent quite a bit of time making games/social WebXR stuff in Three.js. There is unfortunately a ceiling to what commercial success you can have there. Sure there's all the .io games which see a bunch of kids playing games from their school computers, but as much as I'd like to see it, web games aren't going to be as impressive as native PC/Console games. It's not really technology, mostly just the market isn't there.
The models are pretty decent at building simple Three.js/Phaser games, but if you want to work with Unity/Unreal/Godot you're going to need a MCP or other tool to get them to work with the engine's tooling/context. I just so happen to work on one for Unity https://bezi.com
I will say, while I think the current models are very impressive with generating code for most game mechanics. They are still terrible at spatial awareness. Gemini Pro 3.1 is showing some promise here, the latest Opus/Sonnet models are...ok. But there's still a lot left to be desired. You also still really need to know how to make games both creatively and technically to pull off prompting a game into existence.
So are you going to vibecode your way to the next SimCity / Civ without knowing some game dev? Probably not right now and I think that's for the best. People want games that are creative and unique. But a passionate hobbyist who has never made a game, knows some programming, and has a vision for a great game now has an amazing tool set to build their dream game and that's pretty cool!
Shameless plug - and nothing so grandiose as SimCity but I built a pretty substantial 2D/3D blindfold trainer chess game. It's by no means "vibe coded" though, and there's a fair bit of manual work around the 3d modeling that I had to roll myself.
Even with that I'd still say 70% of the code was written using LLMs with the opencode agent.
https://shahkur.specr.net
I wanted to create a game (Pre-AI) and thought it would be easy. Turns out it is not at all.
Sure you might get lucky with the next Angry bird, but there is a whole range of skills required to make a good game, and actually get people to play it.
I even read a couple of books which gave me a better understanding of how out of my depth I was. Books were "Theory of Fun", "Achievement unlocked" and "The art of game design".
AI here appears to be accelerating the ability to see those gaps faster. I think without the understanding of those gaps anything created is going to be lacking.
Having AI generate a game for you no more makes you a game dev than ordering a happy meal makes you a chef.
This is funny as most Michelin star chefs I've had the luxury of knowing love fast food
This 100%. After all, why do you think so many chefs feature "elevated X" items? Have you tried our take on the Taco Bell Chalupa made with A5 wagyu?
Also, they make their kids boxed Mac and Cheese because that's what they ask for.
Gotta love a good false dichotomy on late-night HN.
Call it whatever you want, the game wouldn't exist without them doing something about it. And because they did something about achieving their vision other people get to play the game.
How about we just call these people the game Producers instead. Thats what a producer does anyway right? They make decisions of how the game is built, what goes into it, ect, ect, ect.
Oh please, just stop being silly. Being a game dev means you're building games which means you're thinking about the mechanics of the game and what makes something fun to play. The programming is not "game dev" and offloading it to an AI agent doesn't make someone any less of a game dev.
Well as long as they keep labelling it as AI so I can avoid buying it, we’re all good.
If you buy a game and can't tell it's made with AI, isn't that just as good?
Totally! I'm sick of not being able to see what tools they use. In fact, I feel that we should also start tagging devs that use an IDE instead of properly coding games in notepad.
If you want to see what the HN game dev community is building, I’ve been curating the list here (https://hnarcade.com) for the past months.
The output is quite impressive. And having spoken to a number of the developers, it does seem like AI has had a massive impact on delivering their ideas.
I think the article is being s bit disingenuous. The real problem is triple A developers and publishers pushing for it to replace a not insignificant amount of creative work.
While still over charging everyone and scalpling every $ from everyone with micro transactions and game mechanics that need xp boosters.
I am so bullish on this.
AI is obliterating the barriers to game production for the next generation.
Will this be the next flash revolution? Or is the underlying 'brainrot' actually destructive to creative potential?
I am optimistic about the human spirit in this regard. Making games with AI will be cool when the games are cool, and the only barrier is design.
>Or is the underlying 'brainrot' actually destructive to creative potential?
Creativity comes from constraints. Writing code hasn't been the hard part since the 90s. Deciding which things make for a good game and are worth spending your limited time on is where fun comes from.
AI makes it easy to spit things out, but it doesn't make things fun or good at all.
The most fun games were made when code and resources to code were the constraints.
Time will tell! Certainly AI is unlocking a lot of stuff, a huge number of people who did not have the time or the special skills before to create games will do so. I can think back to my own start in making computer games, using Amos on the Amiga (vastly easier than C or assembly).
It's a separate question whether anything actually good will come out of it. It's incredibly unfair to look at any particular project and say: what, another clone of this or that done idea? Very few things are original in any time. Certainly I didn't make anything particularly original all those years ago. But, soon, there should be really something, if there's really something there. If it's not just burning tokens to copy older ideas. And we'll know it when we see it, this amazing thing that would not have existed otherwise.
Which game engine is recommended for vibe coding games? I assume something like Unity is not very practical, since it's very GUI heavy.
I wish that people would just stop with the "vibe coding" thing.
If you want to get quality results from an LLM use a quality frontier model (I recommend Opus 4.5 thinking) in an agentic Plan -> Agent -> Debug loop inside of Cursor. Roughly 90% of the hate that gets assigned to AI anything is a direct result of the absurd notion that taking the human completely out of the loop is a valuable goal. In reality, it's expensive and almost guaranteed to produce crap.
If you treat LLMs as pair programmers and split your implementation into a set of sequential tasks of a reasonable scope, you can use Unity or Unreal or any number of JS engines built on ThreeJS to produce things that are worth playing.
I would strongly argue that pairing with Opus to write your controller code while you take primary responsibility for interacting with the UI sounds exactly like how you should proceed if you care about the end result.
Im using UE5 which is arguably GUI heavier than Unity. You can get around a lot of the GUI-heavy stuff by using C++ in preference to Blueprint, and/or developing some tools to help you decompile/recompile Blueprints.
It has a pretty cool remote control plugin you can install which can be used to simplify a lot of test cases through automation.
I have a relatively large amount of experience with UE4/UE5 and C++ though, so it's probably not for the absolute beginner or the faint of heart.
I've build 10+ games which I am ashamed to show :)
Straight JS/html/css front-end with zero dependencies works well.
Ask for a node.js backend and can be instantly deployed as client/server or straight to html - multiplayer feels trivial.
C# Monogame works well for something heavier.
You can actually edit Unity scenes directly using the LLM as they're a readable text file which works ok, but Unity is bloatware when you can code it all yourself (it's an absolute nightmare of inexplicable bugs, do not use it. After updating to 0.62f from 0.48f my clang compiler now segfaults while building Webgl - luckily my team mate can do the builds)
The key is building exactly and only what you want and need. Make your design lean, suit the game as you are actually building it not a theoretical overengineered masterpiece - refactors are cheap later, but bloat will kill your project.
I've had my own rollercoaster relationship with Unity over the past decade, but telling people to roll their own game engine so that they can finally make a game is almost universally terrible advice.
People who want to build game engines should build game engines; people who want to build games should absolutely use Unity, Unreal or Godot in no particular order.
It's no different than needing to build a web framework so that you can make a website. The people who do it are often not even aware that they are procrastinating.
Big plus for html/css/js, mostly Pixi 8 or around (I also have a couple threejs). Vanilla JS. I did this by hand before, but having the LLM tweak around the code and styling while I handle more gameplay related things makes this doable (otherwise I just would not have enough free time)
Pick an engine that's been around for a long time like Love2D or PhaserJS as they'll almost assuredly be in the training data of any substantial LLM.
https://github.com/phaserjs/phaser
https://en.wikipedia.org/wiki/Löve_(game_framework)
It's been a while since I used it, but afaik the editor is GUI heavy, the code is still there in C#.
Much like AI is great at Boulter plate code like FE, it's probably great at that sort of Unity code.
> Nobody's sharing their work, because they used AI.
Really. What a stupid statement. As if AI bros aren't screaming all over the internet telling people their vibe-coded projects.
I do use AI for code, but I really don't know why AI users seem to have some sort of victim mentality when there are literally billions of capital stacking on their side.
I strongly disagree. Let me try and change your mind.
Let's say that you have been building your passion project; it could be a game, it could be a synth, it could be a mobile app. You were a year into development already when LLMs blew up, and you very quickly realized that you could leverage them to move much, much faster. This allowed you to tackle a long list of features that realistically would have never left the daydreaming to-do list because in reality, it'd have been a minor miracle to just get to v1.0 much less do all of the cool stuff you thought of along the way. This is the real power of the agentic workflow: in experienced hands, it's an incredible force multiplier.
And yet, even while your churning out features and knocking out bugs that you never would have found on your own without users being upset, the world gets progressively more and more pissy about this moment's Scarlet Letter. Suddenly mentioning that actually, you're getting amazing results from LLMs makes a significant percentage of people irrationally angry, leaving comments that would otherwise be appropriate to shame deviants and cheaters.
Suddenly, you're faced with a really awful realization: you'd love to loudly and proudly show the world how your LLM-assisted dev workflow has allowed you to produce something you honestly might not have ever finished in 2023, but if you do then people are going to write off what you've done as slop before even looking at it, and since you were actually thinking that maybe what you were building might have some legitimate commercial viability, you can't actually advocate for sanity and transparency without taking a serious hit to your product's success. By extension, this means that it might cause real damage to you and your family.
What I am describing is a chilling effect, and it is real. All of the billions in VC have literally nothing to offer someone who just wants to build cool shit and can suddenly do so faster and often better.
At the very least, consider that there is a tangible cost to telling the world that the thing you created exists at least in part because of AI but there is no such thing as people being more likely to pay for something simply because AI was involved. There is only downside. That is the chill.
It's not about victimhood. It is about not fighting battles that can only lose you ground.
i don’t like ai art bc it lacks and sort of “soul” or “human touch” (nebulous and subjective but iykyk) games are just another form of art, so why would i waste my time with just another form of slop?
Similar. I want to see games made by humans who have put in the effort and taken the time to build something good. I don’t want to see the market flooded with low effort AI slop.
If you're going to engage in Glenn Beck Conspiracy Whiteboard Logic on HN, at least start from a point that can be empirically measured.
I mean AI art is for sure super samey and plagiarized and uninteresting stylistically
It's instantly recognizable and makes most indie games that disclose it just dead on arrival
Exactly: https://news.ycombinator.com/item?id=47229458
You should consider that there is a rather large gulf between low-effort crap and using agentic LLMs to make more sophisticated games faster before you downvote me.
It's just not black and white, and to treat it as such devalues the conversation.
I mean all the major hurdles of making a game aren't really helped much by AI...
AI has no training data on complex logic and systems so you gotta do that all yourself.
It definitely doesn't get anything visual right really.
There isn't large amounts of automated testing you can setup ahead of time for a lot of game-play so the AI can't iterate on it to make something work it'll just be hopeless.
The art is also going to all be really derivative plagiarism overly averaged scammy looking stuff. So that's basically an insurmountable hurdle. No unique style.
>I mean all the major hurdles of making a game aren't really helped much by AI...
I have been using Claude Code to develop a game in unreal engine. It is fricking amazing. Its like hiring someone with 10 years experience to work for you. I am really impressed by how it know game patterns.
>It definitely doesn't get anything visual right really. Sometimes it struggles to things things right visually, other times it nails it!
I have been using an MCP from gemini image 1.5 to generate my icons. And once it go my styles down, after 20 experiments, it does really good. Notice: It uses high quality by default which will burn up your credits. But if you turn down the image quality to low, it cost around 3 cents an icon.
>There isn't large amounts of automated testing.
Some things can be easily automated for testing. But other things require play testing.
>The art is also going to all be really derivative plagiarism
I am just using it to generate icons, and it does great. For 3D artwork I either use things from the FAB Store, or I pay a team of artists in Pakistan to do it.
Overall I say it is the equivalent to have a senior dev on your team, for 100 bucks a month
edited for line breaks.
Did you mean gpt-image-1.5? (the Gemini models are Imagen or NB)
this url https://www.tyleo.com/blog gives a 404 when infact it should retrieve a list of all the blog posts. Might want to fix that
It's actually intentional. The literal strings are just routed to pages.
However, if users are trying to do it, it should be a behavior. I'll make a note to change this.
For now, if you'd like to see more posts, you can go to www.tyleo.com and scroll down to the Writing section.
This is a WIP. I hadn't anticipated the popularity my site reached/extent of the blog.