Google Patents To Create AI NPCs That Can Adapt To Every In-Game Situation

The patent wants to add NPCs that do not have limits of the programmed scripts.

By Shahmeer Sarfaraz October 30, 2023

Story Highlights

Google has published a new patent that wants to create NPCs that can adapt to every in-game stimuli on the spot. It argues that the usual scripting methods are quite limiting.
The current NPCs can only react to certain in-game situations they were mapped for. It can break immersion to see NPCs do nothing in some moments.
Google wants to improve NPCs using AI. It proposes using “action grammar” coupled with tokens that are replaced by natural language phrases to define instructions to NPCs.

After the inception of AI tech, video games have started improving at a rapid rate. But in-game NPCs can still feel quite limited in interactions because of their statically programmed nature. Google has recently published a new patent that wants to improve NPCs by making them much smarter and adaptable to every in-game situation that is not already pre-programmed.

Instead of explicitly scripting or mapping stimuli to actions performed by an agent [NPC] in a videogame, natural language phrases in an expression space for the agent are mapped to the actions that are available to the agent, e.g., using a metadata footprint,” states the patent.

The patent dubbed “CONTROLLING AGENTS IN A VIDEO GAME USING SEMANTIC MACHINE LEARNING AND A NATURAL LANGUAGE ACTION GRAMMAR” argues that the current NPCs in games are often scripted and mapped by the devs to react to different stimuli it already expects. These NPCs can be quite dismissive when they are faced with situations that they were not scripted for. The patent wants to solve this dilemma using AI.

The diagram shows a system that controls actions of agents in a game using language processors. — The diagram shows a system that controls the actions of agents in a game using language processors.

When a stimulus has not been explicitly scripted or mapped by the game developer, the videogame either does not respond to the stimulus or responds to the stimulus with a message that indicates that the stimulus has not been understood.”

Moreover, the current method of developing in-game agents requires extensive expertise in various scripting languages from the devs; it can be quite limiting. Google proposes using “action grammar” coupled with tokens that are replaced by natural language phrases to define instructions to NPCs.

Related Content:

The flow diagram shows a way of controlling actions of an agent in a game. — The flow diagram shows a way of controlling the actions of an agent in a game.

The game actions that are implemented for the character are labeled with information indicating a token in the action grammar and the natural language phrases that can replace the token.”

Google wants to include machine learning capabilities with the system to allow for a natural way of interactions between the different NPCs or between NPCs and the player(s). It is worth noting that each character could essentially be given their personality type to deal with specific in-game stimuli in different ways that are more relevant to their behavior.

The diagram implements a semantic natural language processor that controls actions of an NPC. — The diagram implements a semantic natural language processor that controls the actions of an NPC.

The new system will remove the inherent limits in NPCs and allow them to adapt to all sorts of in-game situations that could occur using AI. In other words, it can make the NPCs cleverer and aware of everything going on around them.

Did you find this helpful? Leave feedback below.

Thanks! Do share your feedback with us. ⚡

How can we make this post better? Your help would be appreciated. ✍