Pass the Controller? Communities, Data, and AI in Game Development | GameDev Day 2026
In a Nutshell
Developers must combine quantitative data, direct player observation, and community feedback rather than relying on any single source—players often claim to enjoy features they never actually use. AI can accelerate bug fixing, feedback analysis, and moderation, but game vision, design decisions, and the final "ship it" call must remain human. Early, sustained community building on Discord, Reddit, and store pages is the most reliable way for first-time developers to find and retain their audience.
These notes were generated by AI and may contain inaccuracies.
The panel titled "Pass the Controller" addresses what gamers can tell developers about games and what they want, how data reveals player preferences, and how artificial intelligence is employed in game construction. The discussion also examines situations where community inputs conflict with other data sources, and how developers decide whom to listen to and when to reject opinions. The panel explores how studios collectively learn from these situations and how AI can support newcomers to the industry in learning fundamentals without performing the heavy lifting.
Anton from Berlin is the founder of "The Southern Forest," an independent studio working on "Tideline," a story-driven adventure and action game. He previously worked in mobile gaming at King and Wooga, and supports studios as an expert in game design and development.
Gennady is the CEO and co-founder of General Arcade and co-founder of Beware of Bytes. His studio specializes in porting, re-releasing, and co-developing games across PCs, home consoles, and mobile phones.
Jason is a game developer and educator who shares expertise on YouTube. His background includes MMO games, mobile games, and virtual reality games.
Anthony from Discord is a leading advocate for developers on the platform and has created over 40 indie games focusing on local multiplayer and exploration. His work extends to installation games, art, and live visual performances.
Anton shared that when working on projects with active Discord communities, players frequently understood aspects of games better than the development team. The studio learned from observing player behavior, feedback, and actions. Anton emphasized that designers have assumptions they want to prove, but community feedback always reveals new insights, stating that even the best designer cannot succeed without player feedback.
Anthony explained that Discord communities serve as important extensions of the gaming experience and function as development tools beyond feedback collection. They provide direct communication lines about upcoming content and draw audiences into future developments. Unlike communities existing on external platforms like Reddit, self-hosted communities allow developers to attract and engage their audience directly. Developers have created alternate reality games and game add-ons within Discord servers, customizing them to match game themes and establishing two-way interactions between communities and games.
Anton described retaining quantitative and qualitative data practices from mobile gaming environments like King and Wooga. Quantitative tracking involves monitoring game events and indicators, then combining this with direct player communication. Methods included writing in Facebook communities, joining communities as players, and conducting direct phone calls to players. These conversations extended beyond game topics to understand the broader ecosystem around players. Anton applies these analytical approaches to indie 3D action-adventure development, tracking player progression through specific actions and frequently repeated behaviors to support community observations with data.
Jason noted that in games representing major passion projects, developers often know exactly what they want and how the game should function. External comments can become noise that is difficult to filter for valuable insights. Data becomes crucial for distinguishing useful feedback from the loudest voices on Discord or forums, which may impose poor ideas. However, when developers do not play their own games extensively, direct user interaction and understanding preferences becomes essential.
Jason provided an example from developing a children's game designed for eight-year-olds. Since the developer would not become engrossed in the game, extensive interaction with elementary school children who played daily helped identify what they found fun and interesting. This approach proved effective, though caution is needed when listening to random forum and Discord opinions.
Anthony acknowledged learning from mistakes made as independent game developers, emphasizing that direct audience conversation is more valuable than examining community opinions in public forums.
Gennady explained that General Arcade always examines community feedback for games they work on as co-development partners. This feedback can affect work scope and priorities after client discussion. During technical support for "Project Zomboid," Gennady discovered through Discord and forums that a player had created their own map because they had played extensively and lost the joy of discovering new places. This observation shifted priorities from adding activities to the existing map toward expanding the map to restore discovery joy.
Anton explained the importance of distinguishing between feedback describing problems and suggestions for fixing them. A player might say "This boss is very tough, let's remove it," but investigation might reveal issues with game controls, insufficient save points, or inadequate visual feedback when the boss hits the player. Understanding the purpose of specific game sections allows changing implementation methods while preserving intended outcomes.
Panelists discussed whether studios owe community members who become moderators, whistleblowers, and committees. The consensus was that recognition should be provided without monetary value to avoid legal complications. Common rewards include special badges next to names in forums, early access to content, and occasionally special items or free access. Anton mentioned creating special items for cooperative players at Vanguard, noting that unique items with player names are particularly valued in online multiplayer games.
Anthony described the VIP Players Club in "Jones Journey" and acknowledged that implementing features suggested by professional gamers who provide extensive feedback serves as its own reward. On game creation platforms, ensuring requested features work with existing systems helps creators build desired games.
Anthony recommended mentioning player names in changelogs when they provide feedback or report bugs. Special roles can be granted to frequent contributors whose feedback is implemented, cooperative community members, or long-term bug testers. These roles can persist for the community's lifetime, allowing others to see how individuals helped steer game direction.
Gennady recalled games adding achievements featuring player nicknames and avatars as recognition methods.
The panel addressed situations where player statements contradict tracked game behavior. Anton follows intuition first when facing contradictions, attempting to understand what players are really trying to communicate. Since feedback often comes as solutions rather than problem descriptions, identifying underlying problems leads to worthwhile fixes. In game-making platform development, terminology differences created confusion when users called elements by different names than internal tracking systems.
Anton advised against blindly trusting data without understanding how tracking events were implemented, how graphs were filtered, or how data collection was configured. Teams may discuss decreasing metrics without recognizing underlying code issues causing the apparent problems.
Jason described an MMORPG project where tracking showed high activity completion rates, but surveys revealed players found activities extremely boring. Direct conversation revealed players completed activities only because they were the fastest progression method, highlighting the need to compare multiple information sources.
Jason noted that terminology issues were minimal with younger audiences, but the primary challenge was players claiming to like elements they did not actually engage with. Data analysis revealed players would not provide negative feedback and claimed to like almost everything. In an MMO game, players said they loved all three progression systems, but data showed they only played the adventure system while one system had near-zero player counts. Similarly, children claimed to enjoy mini-games, but data showed they primarily played pet battle modes.
Gennady explained that problems affecting more than 1% of players become high priority depending on severity. Widespread issues require faster intervention than isolated cases. Save file problems warrant investigation even if affecting single players, while graphics issues on single devices might be deferred if potentially caused by user settings or installed software like antivirus programs.
Anthony emphasized determining whether problems are individual or widespread by gathering device specifications and understanding exact circumstances. When patterns emerge in tracking data, issues can be confirmed. Players who agree to be contacted can provide direct problem descriptions and help identify root causes.
To understand the size of a subgroup affected by a problem, developers need to rely on certain data points and build data pipelines within their games and products. This helps determine whether an observation is passing or if more players are affected.
Community building tools depend on the engine being used. Common platforms include Discord servers, Reddit forums, and in-game events.
The discussion covers using artificial intelligence such as large language models to process community feedback and potentially close the feedback loop, including AI adjusting game balance or features based on player opinions.
One developer mentioned a friend who links artificial intelligence to Discord to take feedback from users and work on implementing fixes simultaneously. When users send error reports, the system attempts to reproduce the issue and fix it automatically, then sends a trial version for testing. Most of this automation focuses on technical assets and simple code, which is unlikely to break completely. While releasing versions to the public immediately raises concerns, automatic bug fixing could be interesting for some projects depending on scope and scale.
AI is not just about writing code or creating new features and features. It's about automation in complex processing lines. Some studios use AI when clients allow it, while others have clients who prohibit its use.
When tickets appear on Jira, an AI agent examines the code to find problems and typically returns reports with likely causes and three different ways to fix them. The programmer then decides which fix is appropriate and gives approval to proceed. AI is also used effectively to gather feedback, such as immediately asking ChatGPT for private feedback on performance or questions about the Switch platform when a ported version is released.
GGWP is a company that uses AI for moderation in both communities and games. This is particularly helpful for reviewing large amounts of community feedback. However, AI should be viewed as only part of the bigger picture. Human beings should also look at feedback, talk to people, and maintain presence in the community because AI might miss broader context that a person seeing what's happening throughout the day would catch.
There are concerns about whether AI can pick up on game-specific terminology and truly understand how game features work together. Automated summaries might also overlook important blind spots. The gaming community appears very sensitive to AI use, with some players feeling unappreciated when receiving chatbot responses in customer service. Developers are interested in knowing how far teams go with AI and what red lines they draw to preserve the human element.
When interacting with communities, developers want to sound authentic so people feel their feedback is genuinely valued rather than processed through automated systems. Some developers have found AI-powered chatbots helpful for getting things done and answering questions instantly compared to traditional support experiences. The effectiveness of AI assistants may depend on the size and reach of customer service teams, particularly in MMO games where entire buildings of game managers once helped players.
When asked what decision they would never hand over to community vote, dashboard, or artificial intelligence, panelists gave these responses:
- The decision that a product is complete should remain with the developer who checks everything personally before launch
- Creating the game's vision and the game's technical structure should not be decided by AI
- The game's design, what the game is overall, and what makes it fun and feel right should remain human decisions
- The feeling of the game is considered a purely human thing, as personal style shows in how the game looks and feels
For developers building communities with their first game and no known name, the advice includes:
- Start publishing early and put content online, including videos about the development process
- Determine the target audience and spend time in the communities where those players are located
- Start early and often by getting the community in place while building the prototype
- Ensure the community is on store pages, wishlist features, and any published articles about the game
- The earlier the community is established and the more time spent on social media building space for the game, the greater the chances of finding and attracting the audience
While Discord is a default option, other effective platforms include Reddit and YouTube. Another suggestion is collaborating with other developers who have already created games in the same style or genre to ask for promotional help.
Keep JetBrains in your library
Save the videos and channels worth coming back to, and find them again in one place.
