Quick date note upfront: these comments are from June 2026, not this week, but they've gone genuinely unreported on this site and there's a detail buried in the coverage worth surfacing on its own. CD Projekt Red's joint CEO Michał Nowakowski, currently overseeing The Witcher 4 and the studio behind Cyberpunk 2077,
spoke to Edge's Knowledge newsletter about a conversation that's stuck with him: an AI-first studio founder describing a development pipeline most of the industry would consider unrecognizable.
The Pitch That Gave Him Pause
Nowakowski recounted the founder's pitch almost verbatim: 40 prototypes within a week, narrowed down to five within two weeks, with an actual launch three weeks after that. He said plainly that he knows fully AI-generated games are coming, having spoken directly with people building them, but that he has real doubts about whether that approach is actually the right path to follow. He didn't fully spell out where his skepticism comes from, but pointed to the increasingly brutal competition for player attention, thousands of games releasing every year already, competing against streaming, social media, and everything else fighting for the same hours in a day. A flood of rapid, AI-generated titles adds to that pile rather than solving the actual bottleneck, which he frames as being heard at all, not just being made.
The Genuinely Unreported Detail
Separate reporting on the same conversation surfaced something that hasn't gotten nearly enough attention:
Nowakowski referenced Epic Games integrating models including Claude and Gemini directly into Unreal Engine 6, and said CD Projekt has had unusual early access to explore the engine's internals. If accurate, that's a significant escalation from AI-assisted plugins bolted onto an engine toward AI models built into the engine's actual core tooling, from the company behind the industry's dominant AAA engine. It's the kind of detail that would normally be its own headline, and it's worth watching closely for official confirmation as Unreal Engine 6 gets closer to release.
He's Not Alone in the Skepticism
Nowakowski's doubts were echoed elsewhere in the same reporting cycle. Rufus Kubica, external development director at 11 bit studios, raised a sharper, more practical version of the same concern at Digital Dragons: heavy AI use in early prototypes can create a gap between what a flashy AI-generated pitch promises and whether the actual team behind it has the artists and craft capacity to deliver that quality once real production starts. That's a genuinely useful distinction from the usual AI debate. It's not necessarily about whether AI-assisted prototyping works, it clearly can generate volume fast, it's about whether volume at the prototype stage translates into a studio that can actually finish a polished game once the AI-generated variants stop being the interesting part.
Why Nowakowski's Specific Framing Matters
He made one more point worth sitting with:
large language models are, in his description, fundamentally data aggregators that average out the information they've been trained on, genuinely useful for speeding up routine, well-understood tasks, but not a substitute for actually inventing something new. That's a more precise critique than a blanket "AI can't be creative" take, and it lines up with a distinction we've made on this site before: the difference between AI accelerating execution of a known idea versus AI originating the idea itself.
Coming from someone actively running one of the industry's most respected, craft-focused studios, and who's simultaneously confirming Claude and Gemini are being built directly into next-generation engine tooling, this isn't a dismissal of the technology. It's a specific bet that speed and volume alone won't be what separates a game people actually remember from one that quietly vanishes into that same crowded pile Nowakowski described.
Sourcing note: this piece draws on Edge's original interview as relayed through GamesRadar+, TweakTown, and Outlook Respawn, with the Unreal Engine 6 and 11 bit studios details corroborated separately by Rock Paper Shotgun's reporting on the same Digital Dragons conversation.
Does the "AI accelerates execution, doesn't originate ideas" distinction hold up to you, or is that framing already outdated? Let us know your thoughts in the comments. And if you've got a game you'd like us to review, send it over to editor@gamevibe.io.
