Activision Working On Self-Learning Lifelike NPCs In Games

Activision Publishing has reportedly filed a patent that suggests it is developing a system to enhance the behaviour of non-player characters (NPC) in video games by enabling them to emulate specific human players for diverse in-game functions.

"A new chapter" | Source: Activision Blizzard King


  • Activision Publishing has recently published a patent outlining a system that uses artificial intelligence to enhance non-player characters (NPC) in video games.
  • The system collects data on player behaviour in multiplayer video games, including weapon preferences, engagement frequency, and player interactions.
  • Using advanced machine learning methods, the system generates behaviour patterns for non-player characters (NPC) that replicate real players, making them more lifelike and adaptable.
  • This system can continuously update and modify non-player character (NPC) behaviour in real-time to match changes in human player behaviour, ensuring a dynamic gaming experience.
  • The technology is versatile and can be applied to various video game genres, providing developers with a valuable tool to improve non-player character (NPC) behaviour and enhance player experiences.

Today, we encountered a recently published patent from Activision Publishing titled “Generating Improved Non-Player Characters Using Neural Networks,” filed in June, which details a system employing artificial intelligence to enhance non-player characters (NPCs) in video games.

“The disclosed systems and methods track and continuously monitor data about a player or a multiple players and create a non-playing character (NPC) and/or modify an existing NPC that replicates the player(s) play style,” reads the patent’s abstract.

“The disclosed systems implement an artificial intelligence engine that monitors how a real player responds to one or more events in a game and correlates game outcomes with real player actions, with the actions or reactions of third players, and/or with an amount or extent of engagement. The engine may be used to generate, host, or otherwise provide data representative of one or more NPCs to multiple different games, being hosted by one or more servers, concurrently.”

Schematic illustrating a process of applying a machine learning process to game data and/or human player data. | Source: Patent Public Search

In multiplayer video games, players can play with or against each other. Sometimes, when there aren’t enough human players, non-player characters (NPC) are used to fill the gaps. The current non-player characters (NPC) in video games are often too robotic and lack the creativity and adaptability of human players.

The patent outlines a system that monitors and evaluates data from both human players and in-game events to enable non-player characters (NPCs) to imitate the actions of particular human players, serving various functions such as emulating friends or professional players for practice and entertainment.

The system collects extensive data on how human players behave in a multiplayer video game. This includes data on weapon usage preferences, how often players engage enemies, shot frequencies, time spent in specific areas, and proximity to teammates.

Flowchart illustrating an exemplary process for modifying a non-player character, in accordance with some embodiments of the present specification. | Source: Patent Public Search

Alongside player behaviour data, the system collects details about in-game events and their consequences, encompassing player interactions and their outcomes, with this data subsequently subjected to advanced machine learning methods to facilitate the system in comprehending and deriving insights from human player actions and video game occurrences.

Based on the machine learning analysis, the system generates behaviour patterns for non-player characters (NPC). These patterns replicate the behaviour of real players, making the non-player characters (NPC) more lifelike and adaptable to different gaming scenarios.

What sets this system apart is its ability to continuously update and modify non-player character (NPC) behaviour in real-time. If a human player’s behaviour changes during a video game session, the non-player characters (NPC) will adjust accordingly, ensuring that the gaming experience remains dynamic and engaging.

Additionally, the system can make non-player characters (NPC) perform at a level that matches the performance of a human player in the video game. This ensures that non-player characters (NPC) are neither too easy nor too challenging to interact with.

Moreover, the flexibility of this technology extends beyond the confines of a particular video game genre, platform, or type. Its applicability spans a broad spectrum of video games, demonstrating remarkable versatility in its capacity to enhance the behaviour of non-player characters (NPCs) in various gaming environments.

Whether it’s action-packed shooters, immersive role-playing games (RPG), or strategic simulations, the system’s adaptability makes it a valuable tool for video game developers seeking to provide players with more dynamic and engaging experiences.

While this patent holds significant potential for the video game industry, it’s crucial to keep in mind that it currently exists solely as a patent and does not assure its actual development or integration into multiplayer video games.

Nonetheless, with the incorporation of this technology to produce non-player characters (NPCs) replicating human player behaviours, Activision Publishing is establishing the groundwork for multiplayer gaming experiences that become more immersive and enthralling.

What do you think about this? Do tell us your opinions in the comments below!

Similar Reads: Activision Working On Real-Time Spectator Participation In Games

Was this helpful? 🎮

Good job! Please give your positive feedback ☺️

How could we improve this post? Please Help us. 🤔