AI in game development: An efficient replacement for play-testers

• Author:
• Author name

  Quantumrun Foresight
• July 12, 2022

Artificial intelligence (AI) and machine learning (ML) are being increasingly integrated into game development, where game designers are replacing human play-testers with ML models to fine-tune the development process.

AI in game development context

Internet multiplayer games have grown in popularity since the mid-2000s, enthralling millions of gamers worldwide. However, this success puts pressure on game creators to churn out increasingly well-crafted, bug-free, structured video games. Games can quickly lose popularity if fans and users feel the game is not challenging enough, is not repeatedly playable, or has flaws in its design. 

It typically takes months of playtesting to detect inequities in a newly prototyped game during the game development process. When an error or imbalance is identified, it can take days to alleviate the problem. A recent strategy to combat this issue sees ML tools deployed to change gameplay equilibrium, with ML using its earning algorithms to act as play-testers. An example of a game where this was trialed was the digital card game prototype Chimera, which has been previously used as a testing ground for ML-generated art. The ML-based testing process enables game designers to make a game more interesting, equitable, and consistent with its original concept. The technique also takes less time by running millions of simulation experiments using trained ML agents to conduct research.

Disruptive impact

These ML agents might be used in various ways, including mentoring new players and creating novel playing techniques. Depending on how successful ML is in testing games, developers may consider increasingly deploying it to create games themselves or to reduce their workload. In addition, because these instruments do not comprise any coding, they might provide easy exposure to new game developers by eliminating the need for interacting with scripts. Game design could become more democratized due to this automation, potentially making it more economically feasible to produce educational, scientific, and entertainment games and game-related applications. 

Therefore, AI in game development will potentially make it easier for developers to test-run their games and make improvements in a matter of days, producing complex new games at a much faster rate in the future. According to research, an AI system could estimate game performance based on a forecast and build the whole game using only keyframes and consumer data. In the future, the AI system may be able to create in-game images, sound and even draft the plot on its own.

Implications of AI testing in game development

The wider implications of using AI testing and analysis systems in game development include: 

• Increased speed of game development, allowing companies to increase profits by releasing more games per year.
• Fewer games receiving bad publicity as an AI system would have tested them, and there would be fewer coding errors.   
• Increasing the average duration of games in a variety of genres, as the costs to produce longer storylines and infinite open-world environments will be reduced. 
• The increasing application of games or gamification in non-gaming contexts; for example, brands and marketers may consider more actively creating branded games due to the reduced cost of their development.
• Media companies increasingly diverting a portion of their film and television spending on video game production.

Questions to comment on

• Do types of new gaming experiences may become possible thanks to the AI involvement noted above?
• Share your worst or funniest videogame bug experience.