Blog
Data analyst finds ‘AI stigma’ on Steam can reduce the number of reviews a game gets by around 53%—and the reviews it does get are more negative
Since the advent of generative AI and contemporary chatbots, there’s been endless debate about the ethics of using them in gamedev. But how does AI measure up through a purely pragmatic, business-minded lens? According to a blog from Game Oracle written by market data analyst Ross Burton, stigma around the technology can ward off prospective players.
Game Oracle’s methodology involved taking a sample of 9,879 games released between Jan. and Oct. 2025, “filtering out spam and purely commercial releases,” as well as free-to-play games (granted, this could exclude some relatively popular free-to-play games and those which have been accused of using undisclosed AI art, like FragPunk, which is both). Of the sampled games, 17.9% disclosed AI use.
Taken as a whole, AI use was correlated with slightly less enthusiastic reception: games without AI had slightly more reviews, fewer of them had no reviews at all, and “when focusing on games that received at least 100 reviews,” the median rating was about 4% lower.
However, with the methodology adjusted specifically to compare games that were alike in other ways, things were different. The report states: “After controlling for publisher, developer experience, and game type, developers using AI see a ~53% reduction in reviews compared to those who do not.”
“To explain away the observed penalty, an unmeasured X-factor would need to be strong enough to nearly triple the odds (2.7x) of AI adoption while simultaneously causing a 22% reduction in review counts, independent of publisher backing and developer experience.”
The full breakdown behind these findings is in the report, but it also states that this effect was more pronounced the bigger and more accomplished the developer was. “Our data suggests that for low-quality games, AI makes no difference,” it reads. “But for high-potential games, the ‘AI Stigma’ is real and severely punishes developers who otherwise would have succeeded.”
Games which used AI extensively and remained hugely successful like The Finals, Game Oracle reckons, “highlight the nuance around how AI is used … AI can be used well, or it can be sloppy, and that matters.” The study concludes that “AI is a tool” not to be avoided, but approached cautiously. “Would you avoid using a hammer to build a shed? No, of course not. Just don’t go around hitting everything with it.”
It’s worth noting that plenty of game devs are skeptical of AI’s potential as a mere tool, and things have changed even since 2025. Clair Obscur raked in game of the year awards last year despite its use of AI-generated placeholders, and Crimson Desert has sold millions of copies already this year after doing the same thing—but with the new Crazy Taxi, for instance, it feels as though the developer’s AI use has been talked about more than anything in the game itself.
This is all further complicated by the prevalence of undisclosed AI use, with industry figures like Epic CEO Tim Sweeney pushing back on the very notion of the disclaimers, as well as the ways huge studios continue to invest in the tech. Given that Sony touted AI tools as a means to “unleash the creativity of studios” just a few days ago, I wouldn’t expect these findings to signal an industry-wide shift anytime soon.
