Steam is drowning in AI slop: Here's how Valve plans to fix it

The growing number of AI-generated releases on Steam is not only making it harder for players to find interesting games but also reducing visibility for developers and publishers. Following the flood of AI-generated games at Steam Next Fest (in German), which drew considerable criticism toward Valve, the company is likely well aware of the need to take action. According to recent reports, work on potential solutions is already underway. However, rather than restricting AI-generated games, Valve appears to be focusing on making genuinely good games easier to discover.

The new information comes from Valve employees Tom Giardino and Marc Wilkinson, who answered questions from Korean game developers at Steam Indie Connect in South Korea on October 27, as reported by gaming publication Inven.

Valve acknowledges that this approach can initially give low-quality titles considerable attention. According to Wilkinson, numerous employees are already working on ways to address the issue. Giardino, however, explained that Valve doesn't want to dictate which tools developers should use to create their games. Rather than deciding for itself which games are good or bad, Valve wants to make the platform smarter and improve how players discover titles that match their interests.

Filtering out AI slop without punishing good games

Exactly how Valve intends to achieve this remains unclear. However, the way Steam's current recommendation system works provides some interesting context. As Wilkinson explained, games at Steam Next Fest were displayed largely at random during the first two days to give all participants a fair chance at attracting attention. Only afterward did Steam place greater emphasis on personalized recommendations based on players' interests. This approach can temporarily give low-quality games considerable exposure while worthwhile titles get buried under the sheer number of releases. Valve apparently wants to improve this process. One possible approach would be to incorporate user behavior and reviews into the recommendation algorithm at an earlier stage. However, it remains to be seen exactly how the company will tackle the problem.

Valve faces a classic dilemma here. Mandatory AI disclosures and corresponding filter options could significantly reduce the visibility of AI slop, but they could also end up hurting genuinely good games. After all, the use of generative AI during development doesn't automatically make a game bad. Titles such as Liar's Bar and Crimson Desert might never have reached many players under such restrictions. Opinions are divided on Reddit as well. While some users are calling for mandatory AI labeling and filter options, others would rather judge the quality of new games independently of the tools used to create them.

Feel free to share your thoughts on the topic in the comments below.