The popularity of gaming as an entertainment, training and research tool has fueled an interest in using AI to improve game play. Video games are ideal environments for AI because they provide structure, repetition and reinforcement, making them an excellent environment for algorithms to learn.

꽁머니 is most often applied to Non-Player Characters (NPCs), transforming them into intelligent, adaptive adversaries that mimic human-like behavior. Pathfinding algorithms enable NPCs to navigate complex game environments, while reinforcement learning allows AI opponents to adapt to player actions and create a personalized challenge for each player.

Exploring Procedural Generation in AI Games

More recently, generative AI algorithms have enabled NPCs to speak and move with more naturality. One example is Jitao Zhou’s NPC, which takes speech as input and generates body gestures in real time to create a more realistic character. NPCs with more individualized movement will allow players to better engage and relate to their in-game avatars, improving the overall gaming experience.

NPCs can also be used to train AI systems, providing an opportunity to experiment with different approaches to a problem. For example, a bug in the classic arcade game Q*bert allowed a machine learning algorithm to rack up almost infinite points, and this led to the development of evolutionary strategies. Similarly, the team behind the virtual office software SoWord leveraged game-based AI to develop its agent model of hybrid work.