Study suggests AI is hurting video game sales

Generative AI is now widely used in game development, supporting everything from concept art and translations to coding assistance and internal workflows. On Steam, developers must disclose when AI was used during development. According to a recent analysis, however, that disclosure alone may already be enough to deter many players.

Game Oracle analyzed 9,879 Steam games released between January and October 2025. Free-to-play titles, obvious scam or spam releases and statistical outliers were deliberately excluded. The results at a glance:

Average reviews in the first month after release

• With AI disclosure: 4 (19.8% without reviews)
• Without AI disclosure: 7 (15.2% without reviews)

Average user ratings (at least 100 reviews)
• With AI disclosure: 84.6%
• Without AI disclosure: 88.3%

The gap becomes even more striking when factors such as developer experience, publisher support, genre and release timing are taken into account. According to the study, comparable games with an AI disclosure receive around 52.6% fewer reviews than similar titles without one. Put simply: if a comparable game without an AI disclosure receives 100 reviews, the same game with an AI disclosure would be expected to receive only around 47 or 48. Since Steam does not publish official sales figures, the study uses review counts as a rough indicator of commercial performance.

Game Oracle also stresses that not every relevant factor can be measured cleanly. Budget, marketing, development experience and overall production quality all play a major role. The weaker performance of games with AI disclosures therefore cannot automatically be blamed on AI itself. It is also possible that AI is used more often in projects that are already constrained by tight deadlines, limited budgets or a lack of polish.

Reddit community rejects AI – but not in every case

On Reddit, many users are openly critical of AI in games, often associating AI disclosures with “AI slop,” shovelware and a lack of genuine creative effort. The argument is simple: if developers appear to be cutting corners, players see little reason to spend money on the game. For many, AI therefore seems less like a sign of innovation and more like evidence of cost-cutting. Others push back against the idea that AI automatically ruins game sales. A common counterargument is that bad games simply sell poorly, and that many titles in the study may be small, low-visibility or poorly executed projects regardless of AI use.

For much of the community, the type of AI use also makes a difference. AI-assisted coding, debugging and temporary placeholder assets are generally viewed more favorably than AI-generated artwork, cosmetics or synthetic voices. In other words, many players appear willing to accept AI as a supporting tool, but become far more critical when it seems to replace human creativity.

For developers, the takeaway is clear: AI can reduce some of the workload, but it cannot replace creative vision or good game design. Studios that use it merely to push low-effort content onto Steam are likely to face growing resistance. Used carefully, sensibly and without a visible loss of quality, however, AI does not necessarily have to become a liability.