Generative AI has spent the past few years entering the games industry like a consultant who was invited to one meeting and somehow ended up with access to the entire building. It has been presented as a solution for concept art, code, dialogue, localisation, testing, marketing, research, prototyping, and almost every other part of development that involves time, money, or a person who would ideally like to remain employed. The argument is usually framed around efficiency: games are expensive, teams are under pressure, and generative AI might help people produce more with less. Now, a growing group of studios is making the opposite decision.
Aftermath has begun maintaining a living list of game studios that refuse to use generative AI. The list includes tiny independent teams, established studios, and companies behind recognisable games such as Warframe, The Long Dark, Tunic, and Goodbye Volcano High. Their reasons vary, but the same concerns appear repeatedly: copyright, labour, environmental cost, unreliable output, legal uncertainty, creative ownership, and a simple desire to make the work themselves. This is easy to frame as an argument about technology, but it is more interesting as an argument about design. When a studio promises that its game is human-made, it isn’t only describing the production pipeline. It is telling players what kind of relationship they are being asked to have with the work, making authorship, accountability, and creative intent part of the product.
The Refusal Is A Production Decision
A no-generative-AI policy can sound ideological because, for many studios, it is. The technology raises serious questions about how training data was obtained, whose work was used, who was compensated, and what happens to creative labour when imitation becomes cheaper than collaboration. However, the studios featured by Aftermath also describe practical production problems. Generated output still needs to be checked, code still needs to be understood, art still needs to fit a visual language, and writing still needs to support characters, tone, pacing, and world rules. Legal ownership needs to be clear and contractors need consistent expectations. A tool that produces something quickly can still create slower problems later if nobody understands where the output came from, why it works, or what it quietly copied on the way through.
The Efficiency Promise vs The Production Risk
| Generative AI Promise | Production Risk |
|---|---|
| Faster asset creation | More time checking consistency and provenance |
| Cheaper content | Hidden legal, reputational, and revision costs |
| More ideas | More undirected material competing for attention |
| Less specialist dependency | Fewer specialists available to judge the output |
| Rapid prototyping | Temporary work quietly becoming final work |
| Scalable production | A larger volume of decisions nobody fully owns |
Efficiency isn’t simply the speed at which something appears. It is the speed at which a team can make a reliable decision and carry that decision through the rest of production.
Insider Tip: A tool hasn’t saved time if the team spends the rest of the project repairing decisions nobody remembers making.
Human-Made Is Becoming A Design Promise
Players have always cared about who made games. Studio names, directors, composers, artists, modders, performers, and individual developers become part of how communities understand the work. People follow Supergiant, Kojima Productions, Remedy, Larian, FromSoftware, and countless smaller teams because a studio name creates an expectation about taste. Generative AI makes that relationship more visible because it introduces uncertainty about where creative decisions came from. Was the character designed by an artist? Was the dialogue written by someone who understood the scene? Was the voice performed with consent? Was the code built by a programmer who can maintain it? Was the store art representative of the game, or was it generated five minutes before the page went live because somebody remembered the thumbnail was due? A human-made promise gives players a clearer answer. It says that there are people behind the decisions and that those people are willing to be associated with the result.
Feature Promise vs Authorship Promise
| Traditional Feature Promise | Human-Made Promise |
|---|---|
| The game contains a large world | People deliberately shaped that world |
| The game has thousands of lines of dialogue | Writers are accountable for the voices and meaning |
| The game offers unique art direction | Artists made and defended aesthetic choices |
| The game contains complex systems | Developers understand how those systems work |
| The game will be supported after launch | A team can explain, maintain, and revise its work |
This doesn’t mean every human decision is brilliant. Humans gave us horse armour, escort missions, and inventory limits that somehow allow seventeen swords but not one additional mushroom. Human authorship isn’t a guarantee of quality, but it is a guarantee of responsibility.
Insider Tip: Players don’t need every creative decision to be perfect. They need to believe that somebody actually made it and can explain why it belongs.
Creative Friction Is Not Waste
The strongest argument for generative AI is often that it removes friction. The artist gets concepts faster, the writer gets variations faster, the programmer gets a code suggestion faster, and the designer gets fifty mechanic ideas before their coffee has achieved basic consciousness. Some friction should be removed. Nobody becomes a better designer by manually renaming four thousand files or spending a week copying values between spreadsheets because two tools refuse to speak to each other. Creative friction is different. The struggle to solve a visual problem can produce an art direction, the disagreement over a character can expose what the story is really about, and the failed prototype can teach the team which part of the fantasy matters. The programmer who traces a bug through the system learns relationships that later help them build something better. These moments look inefficient when measured as output per hour, but they become extremely valuable when measured as understanding.
Wasteful Friction vs Creative Friction
| Wasteful Friction | Creative Friction |
|---|---|
| Repeating a solved administrative task | Discovering what the game should feel like |
| Moving the same information between tools | Testing whether systems support the fantasy |
| Rebuilding an asset because the brief was unclear | Iterating because the first idea wasn’t strong enough |
| Waiting for an avoidable approval chain | Debating a decision that affects the whole experience |
| Fixing preventable pipeline failures | Learning from a prototype that revealed the wrong direction |
The danger appears when every difficult creative task is treated as a bottleneck. If the goal is only to make the difficult part disappear, the team may also remove the thinking that gives the game its identity.
Insider Tip: Automate repetition. Be careful about automating the moment where the team discovers what it actually believes.
The Process Teaches The Team
Game development doesn’t only produce a game. It produces capability. Every completed environment teaches the art team something about composition, readability, performance, and mood. Every quest teaches the design and narrative teams how the game responds to player behaviour, while every system teaches programmers where the architecture bends and where it breaks. That knowledge compounds across the project and, if the studio is stable enough, across multiple games.
Generative tools can produce an answer without transferring the reasoning behind it. That isn’t always a problem. Developers use libraries, engines, middleware, and existing tools constantly; nobody needs to forge their own graphics card before they are permitted to render a triangle. The difference is whether the team retains enough understanding to judge, adapt, and maintain the result. If a junior developer receives an answer before they understand the question, the task may be completed while the skill remains undeveloped. If a studio repeatedly replaces practice with generation, it can become faster at producing material while becoming weaker at making decisions. That isn’t efficiency. It is capability debt.
Insider Tip: If a tool completes the task but leaves the team less able to solve the next problem, part of the cost has been pushed into the future.
Players Are Reading The Production Process
The industry has trained players to examine how games are made. Early access, developer diaries, crowdfunding, labour reporting, union campaigns, behind-the-scenes documentaries, public roadmaps, and social media have all made production more visible. Generative AI adds another layer to that visibility. Players now inspect key art, dialogue, voices, textures, promotional images, and even patch notes for signs that something was generated. Sometimes those suspicions are correct, while at other times human artists are accused of using AI because their work contains a hand with an unusual pose or a texture that looks slightly too smooth.
That uncertainty can damage the very people a no-AI position is meant to support. The useful response isn’t a purity hunt where every creative choice becomes evidence in a small internet court. It is transparency. Studios need to define what they mean, disclose the boundaries, and keep enough production history to support their claims.
Unclear Provenance vs Transparent Provenance
| Unclear Production | Transparent Production |
|---|---|
| Players guess whether assets were generated | The studio explains its policy |
| Human artists are forced to prove themselves | Credits and process support authorship |
| Contractors work under different assumptions | Contracts define the same expectations |
| Marketing makes broad claims | The studio states specific boundaries |
| Mistakes become public scandals | Problems can be traced and corrected |
Transparency doesn’t require publishing every sketch, commit, and awkward placeholder voice line. It means the studio can tell players what it did, what it didn’t do, and who is accountable when the policy fails.
Insider Tip: A trustworthy production claim needs evidence, not just a badge on the store page.
A Policy Has To Survive Production
Saying “we don’t use generative AI” is easy at the announcement stage. Keeping that promise across a multi-year production is harder. Modern games depend on contractors, outsourcing studios, middleware, asset stores, localisation companies, marketing agencies, platform tools, analytics services, and software that may add generative features without anyone asking. A studio can reject AI-generated final art while allowing code assistance, reject public models while using a private model trained on its own data, or ban generated content in the game while using it for internal ideation. Those are different policies, and players shouldn’t be expected to guess which one applies.
Public Commitment vs Operational Requirement
| Public Commitment | Operational Requirement |
|---|---|
| “Our art is human-made” | Define rules for concepts, textures, UI, marketing, and contractors |
| “Our writing isn’t generated” | Cover dialogue, lore, localisation, store copy, and community posts |
| “Our code is human-written” | Define rules for autocomplete, assistants, plugins, and outsourced engineering |
| “Our voices are human” | Record consent, usage rights, and restrictions on model training |
| “Our game is AI-free” | Document exactly which technologies the term includes |
A serious policy needs ownership. Someone must answer questions, review tools, update contracts, track exceptions, and respond when a third-party asset enters the project without clear provenance. Otherwise, “AI-free” risks becoming the new “handcrafted”: a lovely word stretched across so many production methods that it eventually means “please don’t look too closely”.
Insider Tip: If nobody owns the policy, production pressure will rewrite it one exception at a time.
AI-Free Doesn’t Automatically Mean Good
There is an important limit to this argument. A studio can avoid generative AI and still make a bad game. It can underpay people, mismanage production, release unfinished work, ignore accessibility, mistreat contractors, and produce an experience with all the emotional depth of a software licence agreement. Human-made isn’t a substitute for good design.
It is also possible for developers to use generative tools thoughtfully, disclose that use, work with ethically sourced systems, and retain clear human responsibility for the outcome. The useful question isn’t whether technology touched the project because games have always been shaped by tools. The question is what the tool replaced, what it enabled, what risks it introduced, and whether the team and audience understood that exchange. The studios on Aftermath’s list are drawing a clear boundary because they believe the exchange isn’t worth it. That clarity is valuable even for studios that draw the boundary somewhere else because it forces the industry to stop talking about AI as an inevitable weather system and start treating it as a design and production choice.
Insider Tip: “Everyone will use it eventually” isn’t a strategy. It is a way to avoid explaining what the tool is actually for.
The Design Lesson For Studios
The larger lesson isn’t that every studio must adopt the same policy. It is that production values are becoming part of player-facing design. Studios considering generative AI should be able to answer a few basic questions:
- What problem does this tool solve that the existing team and pipeline can’t solve responsibly?
- Which parts of the creative process must remain attributable to specific people?
- How will generated material be checked for quality, rights, bias, and consistency?
- What happens to junior development and team capability if the task is automated?
- Do contractors and external partners follow the same rules?
- How will the studio explain its use-or refusal-to players without hiding behind vague language?
- Who remains accountable for the final decision?
These aren’t public-relations questions added after production. They shape hiring, contracts, pipelines, art direction, documentation, team culture, and the kind of game the studio becomes capable of making. The tool is part of the system, the policy around the tool is part of the system, and the trust created or destroyed by that policy is part of the player experience.
Final Thoughts
Aftermath’s list matters because it rejects the idea that generative AI adoption is automatic. These studios aren’t waiting for the industry to decide what the future looks like. They are defining the kind of production culture they want and asking players to judge the work alongside that promise. For some players, “human-made” will become a meaningful reason to buy. For others, the final quality of the game will remain the only thing that matters. Most will probably sit somewhere in between, caring less about abstract declarations than whether the studio is honest, the work feels intentional, and the people behind it were treated with dignity. That is the real design consequence.
Generative AI is usually sold as a way to make more content. The studios refusing it are arguing that the value of a game isn’t only the amount of content it contains. The value also lives in the choices, relationships, skills, disagreements, mistakes, and strange human decisions that made the experience distinct. Human-made doesn’t guarantee a great game, but it makes a clear promise: people made these decisions, people stand behind them, and people can be held responsible for what the game becomes. That promise may end up being more valuable than another thousand generated assets nobody remembers asking for.
That’s it for this one! Subscribe to The Design Lab for more breakdowns and analysis. Please like, share, and comment if you found this article useful AND…
Need help applying these concepts to your game? Most games don’t fail because of bad ideas. They fail because the systems don’t work together to evoke immersion.
> Work with me:
https://thedesignlab.blog/services/
> Or go deeper into this framework:
Register for my upcoming book – Immersive Design Framework – below and get a second book, now, free!
Grab Myles’ FREE ebook now and find 15 indispensable design patterns that will equip you to craft exceptional Web3 gaming experiences. We’ll also notify you when his new book on immersive design is out!