High-quality generation of dynamic game content via small language models: A proof of concept

2026-01-30

AI Summary

This work addresses the challenges faced by large language models in dynamic game content generation—namely narrative incoherence, high computational cost, and cloud dependency—while acknowledging that small language models, despite their suitability for local deployment, often suffer from insufficient output quality. To overcome these limitations, the authors propose training specialized small models through highly structured task constraints and aggressive fine-tuning, leveraging domain-specific data synthesized via directed acyclic graphs (DAGs). The approach is validated in the context of minimizing reputation-based argument cycles in role-playing games, demonstrating predictable latency, real-time performance compatible with standard game engines, and a clear trade-off between task scope and model specialization. This establishes a practical foundation for deploying modular intelligent agents in interactive narrative systems.

Technology Category

• Humans and AI: Game Design — Procedural Content Generation & Storytelling
• Natural Language Processing: (Large) Language Models
• Multiagent Systems: Adversarial Agents

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Large language models (LLMs) offer promise for dynamic game content generation, but they face critical barriers, including narrative incoherence and high operational costs. Due to their large size, they are often accessed in the cloud, limiting their application in offline games. Many of these practical issues are solved by pivoting to small language models (SLMs), but existing studies using SLMs have resulted in poor output quality. We propose a strategy of achieving high-quality SLM generation through aggressive fine-tuning on deliberately scoped tasks with narrow context, constrained structure, or both. In short, more difficult tasks require narrower scope and higher specialization to the training corpus. Training data is synthetically generated via a DAG-based approach, grounding models in the specific game world. Such models can form the basis for agentic networks designed around the narratological framework at hand, representing a more practical and robust solution than cloud-dependent LLMs. To validate this approach, we present a proof-of-concept focusing on a single specialized SLM as the fundamental building block. We introduce a minimal RPG loop revolving around rhetorical battles of reputations, powered by this model. We demonstrate that a simple retry-until-success strategy reaches adequate quality (as defined by an LLM-as-a-judge scheme) with predictable latency suitable for real-time generation. While local quality assessment remains an open question, our results demonstrate feasibility for real-time generation under typical game engine constraints.

Problem

• dynamic game content generation
• small language models
• offline games
• narrative coherence
• real-time generation

Innovation

• small language models
• dynamic game content generation
• task-scoped fine-tuning
• synthetic data generation
• real-time narrative generation