While AI and gaming have a decades-long history, developers are struggling to integrate this technology into modern games. They run into five common issues, such as disclosing AI-generated content, balancing performance and quality, keeping gameplay fair and fun, and more. Here is how developers can address these challenges.
SwissCognitive Guest Blogger: Zachary Amos – “5 Challenges of Implementing AI in Gaming”
Artificial intelligence has almost always been a fundamental part of video games, from computer chess to first-person shooters. However, people now associate the term with advanced machine learning models and generative AI, not traditional behavior trees or pathfinding algorithms. It faces several challenges when implemented in gaming.
How Do Traditional and Modern AI Differ?
Among AI’s hundreds of thousands of applications, its use cases in the gaming industry are among the most prolific. Everything from side-scrolling platformers to top-down strategy games has numerous features controlled by this technology.
Traditional AI is based on predefined, rule-based systems. Developers mainly use it to dictate nonplayable character (NPC) behavior. For example, the Goombas in Super Mario Bros. use a behavior tree and a finite state machine to move in one direction until they encounter a wall, a pipe or another enemy.
A finite state machine is a computational model that can only be in one of a limited number of states at any given moment. It transitions between them based on player input. Game developers can use it to trigger animations and movements related to idle, walking and attacking modes.
Although this technology has been popular since the earliest days of computer programming, it is slowly being phased out by modern AI. Since machine learning models can adapt to new information, they can learn from player interactions. This enables them to respond dynamically to player input.
Examples of Advanced AI in Video Games
Generative AI is becoming a common tool for asset creation. Game developers describe what they want in plain language, and the algorithm delivers. Machine learning models are less common because they take time, resources and knowledge to build. However, several games already use this technology.
In Dragon’s Dogma 2, the followers, referred to as pawns, learn from the player. Their play style constantly adapts based on gamers’ actions and the information they learn about the world through exploration. This self-learning AI influences how they fight, collect resources and help allies. Even their voice lines change.
Some developers are using AI to turn formerly strictly single-player games into multiplayer experiences. The puzzle-platformer Little Nightmares III features two-player online co-op, a first in the franchise. If fans play alone, the second character is controlled by an AI companion. It responds to player input to help solve complex challenges.
Challenges of Implementing AI in Gaming
While AI and gaming have a decades-long history, developers are struggling to integrate this technology into modern games. Whether they use generative AI to create in-game assets or build machine learning models to dictate NPC behavior, they run into the same five issues.
Disclosing AI-Generated Content
In 2025, Activision only admitted to using generative AI to create some Call of Duty: Black Ops 6 assets after users called it out. Players had noticed that a zombie Santa loading screen had six fingers, a telltale sign of AI. Some of the assets players suspected were AI-generated were part of paid bundles. One of them cost 1,500 COD points, equivalent to $15.
Some companies only use AI for conceptualization, while others use it to create almost everything. While digital storefronts like Steam now require publishers to disclose the use of generative AI if it significantly impacts the final product, there are no standardized rules.
Balancing Performance and Quality
AI is resource-intensive and can strain hardware. For example, while Nvidia’s DLSS reduces the relative load on the graphics processing unit by rendering at a lower resolution, it increases the central processing unit (CPU) load by generating more frames. Balancing performance and quality is challenging, especially when AI is a game’s main component.
Keeping Gameplay Fair and Fun
Games typically aim for repeatability, predictability and performance, as lifelike behavior is difficult to script. Also, players may not like enemies that think like humans because it makes combat and stealth considerably harder.
Overshadowing the Art Direction
A game designer architects the gameplay, worldbuilding and mechanics, while an artist creates the characters, props, environments and textures. They might spend dozens or hundreds of hours on a single project. What happens when generative AI ruins their vision?
Nvidia faced backlash from gamers after showing how its DLSS 5 tool could overhaul graphics. It radically changed characters’ appearances. For example, it gave Grace from Resident Evil: Requiem airbrushed makeup and highlights. While the AI upgrade looked photo-realistic, some fans were concerned it could overshadow the game’s art direction.
Updating Models Post-Launch
Thanks to digital downloads and widespread internet access, patching games post-launch is common. However, that doesn’t mean updating machine learning models is easy. Retraining AI is a resource-intensive, time-consuming process. Also, significantly changing model behavior midgame could introduce game-breaking bugs or performance issues.
What Will the Future of AI in Gaming Hold?
AI’s role in gaming is growing. It can streamline development by offering worldbuilding ideas, identifying lore contradictions and creating character concepts. During gameplay, it can optimize level difficulty, personalize players’ experiences or offer support.
Since this technology is so promising, these development challenges likely won’t hold it back. However, that doesn’t mean gaming companies shouldn’t address them. Solving issues both big and small helps foster acceptance among skeptics and pushes developers to create bigger and better worlds for players.
About the Author:
Zac Amos is the Features Editor at ReHack, where he writes about artificial intelligence, cybersecurity and other tech topics.
Der Beitrag 5 Challenges of Implementing AI in Gaming erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.
