For the first time, most of the people who actually make games think generative AI is making their industry worse. In the 2026 State of the Game Industry survey, run by the organisers of the GDC Festival of Gaming, 52% of professionals said generative AI is having a negative impact on the field. A year earlier that figure was 30%. Two years ago it was 18%. A view that used to belong to a vocal minority is now the majority position, and it hardened while AI tools were spreading through studios rather than retreating from them.
The survey drew on more than 2,300 industry professionals, and the disapproval ran deepest among the people closest to the craft. Those working in visual and technical art (64%) and game design and narrative (63%) registered the strongest opposition, with programmers close behind. Only around 7% of respondents said AI was having a positive effect, down from 13% a year earlier, and even that thin support skewed toward executives rather than the people shipping the work. One anonymous respondent told the survey they would “rather quit the industry than use generative AI.”
Part of what is driving the shift is visible on the storefronts themselves. The cost of producing a passable-looking game has fallen far enough that open marketplaces are absorbing a wave of low-effort, AI-generated titles, a phenomenon developers have started calling “gameslop.” Valve has responded by requiring AI disclosures on Steam store pages, and by March 2026 more than 7,300 games carried that label. What Valve has not done is turn those games away. Its stance is to label AI content rather than restrict it, which leaves the work of discovery sitting with the storefront and the player.
That hands-off approach has a logic to it. A platform that declines to make taste judgements can grow its catalogue almost without limit, and open submission is how storefronts from Steam to itch.io built their libraries in the first place. The difficulty the GDC numbers point to is that volume and quality have started to pull against each other. When nothing is turned away the catalogue keeps expanding, but the typical game gets weaker, and the player has to dig through more of them to reach anything worth the time.
A smaller set of platforms is making the opposite bet, treating curation as the product rather than an overhead. The browser-gaming platform Poki reviews every game by hand and publishes roughly one new title a day, holding its catalogue of more than 1,500 games rather than opening the door to anyone. The constraint has not capped its reach: the platform passed 100 million monthly players in 2025. That last detail is the one worth sitting with, because it cuts against the assumption that hand-curation is a boutique strategy that cannot reach real scale.
None of this makes the open storefronts wrong to stay open, or curation a complete answer to discoverability. A hand-reviewed catalogue is only as good as the people doing the reviewing, and a closed door keeps out plenty of worthwhile work along with the noise. What the survey captures is something narrower and harder to argue with: the workforce has stopped accepting that more games automatically means better games, at exactly the moment the tooling has made producing more games close to effortless.
For players, the practical effect is that choosing what to play is becoming more work. Every hour a marketplace spends not sorting its catalogue is an hour the audience spends sorting it instead, scrolling past near-identical titles to find the one that was worth making. Platforms that take that job on themselves are, in effect, selling a filtered shelf, and the GDC sentiment suggests a growing share of players would value one.
The 2026 report is not evidence that generative AI is leaving game development. More than a third of professionals already use it, mostly for research and writing code rather than finished art, and that share is unlikely to shrink. The line being drawn is about where it belongs: AI as a tool inside a careful process is one thing, and AI as a replacement for the process is another. For the storefronts, the 52% figure is a warning that players may soon start drawing the same line, and rewarding whoever did the sorting for them.
Leave a Reply