We’ve been circling this argument for years, and now CD Projekt Red’s co-CEO has poured a little gasoline on it
Michał Nowakowski says fully AI-generated games are on the way. That’s the headline, but the more interesting part is the tension inside his comment: he thinks the tech could produce something decent, yet he does not believe that should be the industry’s destination.
Nowakowski told Edge’s Knowledge newsletter that he has spoken with AI-based studios, and some of them believe they can move at a speed that would have sounded absurd not long ago. One studio owner reportedly told him they can make 40 prototypes in a week and even put together a full game in that same window. That is the sort of claim that makes us sit up straight, then immediately ask the obvious follow-up question, which is: what exactly counts as a game here?
What Nowakowski actually said
This frames Nowakowski’s comments as a warning, not an endorsement. His position is basically that fully AI-generated games may arrive sooner than many people expect, but he doubts that this is where game development should be headed.
That distinction matters. A lot of the AI conversation in games gets flattened into two camps, with one side pretending the machines will replace everyone by Tuesday and the other pretending AI has no use at all. Reality, as usual, is messier. Tools that speed up certain parts of development already exist, but a fully AI-generated game is a much bigger claim than using AI for concept work, QA support, localization, or placeholder assets.
Nowakowski thinks AI can accelerate parts of development, but cannot replicate what humans do. That is not a new argument, but it is a useful one, because it gets to the real fault line. The question is not only whether AI can generate content. The question is whether it can generate the kind of intent, cohesion, and taste that make a game feel like it was actually made by someone who understood what they were building.
Why this debate keeps coming back
We keep running into the same issue because game development is made of a lot of different jobs, and AI touches them unevenly. It is one thing to have a system spit out mockups, draft text, or rough prototypes. It is another thing entirely to trust that same system with pacing, mechanical balance, readable art direction, or the kind of hand-built design that gives a game its identity.
That is why the conversation around AI in games has stayed so heated. Studios want faster iteration. Publishers want lower costs. Developers want tools that remove busywork. Players, meanwhile, mostly want finished games that feel coherent and worth their time. Those goals overlap in some places, then clash hard in others.
Nowakowski’s comments land in that gap. He is not claiming the technology is useless. He is saying that even if it can produce something functional, that does not automatically make it a model we should celebrate.
The practical difference between AI assistance and AI-authored games
This is where we should be precise, because the terminology gets sloppy fast. There is a real difference between using AI as part of a production pipeline and handing the entire pipeline over to it.
| Approach | What it means | Why it matters |
|---|---|---|
| AI-assisted development | Using AI for narrow tasks like prototypes, asset support, or workflow speed | Can reduce repetitive work without replacing full creative direction |
| Fully AI-generated games | AI produces the game itself with little or no human authorship | Raises bigger questions about quality, consistency, and creative intent |
That table is the heart of it. Nobody serious is pretending AI cannot do anything useful. The concern is what happens when the useful part turns into the whole product. If a studio can crank out 40 prototypes in a week, that may sound efficient, but prototypes are cheap for a reason. The hard part is choosing which ideas deserve to become a real game, then shaping them until they hold together.
And that is before we even get to the very human problem of taste. Plenty of games fail because they are technically functional and spiritually empty. AI can optimize for output. It cannot, at least not in any way we can responsibly pretend is solved, guarantee that a game has point of view.
Why players are skeptical
Fans are already wary of generative AI assets, and that tracks with the broader mood around the medium. Players notice when something feels off. We are very good at spotting visual repetition, awkward animation, strange UI copy, and the general cheapness that comes from content assembled without enough human oversight.
- Players want consistency, not just volume.
- They want intention, not a pile of generated assets.
- They want value, which is hard to sell if the result feels disposable.
That is the big risk here. A fully AI-generated game might technically exist, and it might even be competent in some narrow sense. But competence is not enough to win trust. We have all played enough forgettable games to know that an efficient production pipeline does not magically create something worth caring about.
What this likely means for the industry
For now, the safest read is not that AI games will replace traditional development, but that the pressure to use AI will keep growing. The moment one studio says it can prototype at a ridiculous pace, everyone else in the room starts feeling that pressure too. That is how these things spread, not through grand philosophical debates, but through competitive anxiety.
Still, Nowakowski’s caution is worth taking seriously because it comes from inside the industry, not from a comment section war. He is acknowledging the speed advantage while also pushing back on the idea that speed is the whole story. That is probably the most grounded stance we are going to get right now.
Nowakowski thinks fully AI-generated games may arrive soon, but he does not see that as the future we should want. That feels like the cleanest summary. The tech may keep moving. The argument about what game development should be, though, is far from settled, and we are all still stuck in it together.
What we should be watching next
If we are trying to read the tea leaves here, the important thing is not whether one fully AI-generated game appears first. The real story is how quickly studios start treating AI output as a normal part of production, then how openly they describe that process to players.
That is where the trust issue lives. If AI becomes a behind-the-scenes tool that helps people build better games, most players will probably accept that just fine. If it becomes a way to flood the market with content that feels assembled instead of made, the backlash will be brutal, and honestly, earned.
So yes, fully AI-generated games may be coming. The harder question is whether we’ll actually want to play them once they get here.
