NPC (Non-Player Character)
What is NPC (Non-Player Character)?
An NPC is any character in a game or virtual world that's controlled by the computer rather than a human. AI is making NPCs far more intelligent, giving them the ability to hold natural conversations, remember interactions, and behave in believable, unpredictable ways.
At a glance
- Also known as
- Non-player characterAI characterBotVirtual agent
- Used for
- Populating game worldsDelivering narrative and dialogueProviding opposition or assistance to playersVirtual extras in film and immersive experiences
- Common tools
- Unreal engineUnityInworld AIConvaiCharacter.AI
- Related terms
- Game assetProcedural generationGenerative AIVirtual productionBehaviour tree
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
How it compares
A player character is controlled directly by a human, with the player's inputs translating into on-screen actions. An NPC is controlled entirely by the game's software, following scripted rules or AI-driven behaviour. The distinction is foundational to game design, though the line is blurring in experiences where AI characters can initiate and sustain conversation indistinguishable from human players.
Think of it like…
An NPC is like a theatre actor performing a supporting role in an improvisation show: they follow a general script and set of motivations, but must respond in real time to whatever the lead performer (the player) throws at them. The best NPCs, like the best supporting improvisers, make the world feel real by reacting convincingly to the unexpected.
Pro tip
When integrating LLM-powered NPCs into interactive experiences, define a detailed character system prompt that specifies the NPC's personality, knowledge limits, and conversational boundaries before deployment: without these constraints, LLMs can produce responses that break immersion, reveal technical details, or step outside the narrative world entirely.
Types and variations
- NPCs can be categorised by their role and complexity.
- Ambient NPCs exist purely to populate the world ( pedestrians, wildlife, background crowd members ) and typically use simple looping behaviours.
- Quest or story NPCs are central to narrative progression and carry scripted dialogue and triggered events.
- Enemy or adversarial NPCs use pathfinding and combat AI to challenge the player.
- Companion NPCs follow and assist the player, requiring more sophisticated awareness of player intent.
- At the frontier, conversational NPCs powered by LLMs can engage in unscripted dialogue and form dynamic relationships: a category that is rapidly moving from research into commercial game and experience production.
Ready to make your first scene in Morphic?
Try MorphicCommon use cases
- NPCs are essential to virtually all game genres and interactive experiences.
- In open-world games they provide the ambient population that makes environments feel inhabited.
- In narrative games they carry plot, emotion, and moral weight.
- In virtual production and immersive entertainment, AI-driven NPC characters serve as interactive guides, narrative companions, or autonomous virtual cast members.
- Increasingly, NPC technology is being applied to training simulations ( medical, military, and corporate ) where realistic human behaviour in a safe, repeatable virtual environment provides valuable learning scenarios.
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
FAQs
NPCs use pathfinding algorithms ( most commonly A* (A-star) ) combined with a navigation mesh (navmesh) that defines the walkable surface of the environment. The game engine calculates a path from the NPC's current position to its target, steering around obstacles in real time.
Traditionally scripted NPCs can only respond to situations their designers explicitly anticipated. AI-driven NPCs ( particularly those using LLMs ) can respond to novel inputs, hold contextually coherent conversations, and adapt their behaviour based on accumulated interaction history, creating experiences that feel genuinely responsive rather than canned.
They are beginning to be used in commercial products, but challenges remain around latency, cost per inference, consistency of character, and preventing outputs that break narrative immersion or game rules. Hybrid approaches: combining LLMs for dialogue with rule-based systems for behaviour and safety: are the current practical standard.
In virtual production, NPC-like autonomous characters can populate digital sets as background cast, respond to directorial instructions in real time, and provide dynamic crowd or ambient life that would be prohibitively expensive to animate manually. Real-time game engines enable directors to see and adjust this behaviour interactively on set.
In traditional games, persistent NPC change is scripted through trigger conditions rather than genuine learning. AI-driven NPCs with memory systems ( storing summaries of past interactions ) can adapt their behaviour and attitude over time based on accumulated experience with the player, creating a form of dynamic relationship that persists across sessions.
Key concerns include ensuring NPCs cannot be manipulated into producing harmful content, managing player attachment to AI characters who simulate emotional connection, the potential for NPC voices and likenesses to infringe on real performers' rights, and transparency about when a player is interacting with an AI rather than a human in shared online environments.
