Word2World: Generating Stories and Worlds through Large Language Models
This work addresses the problem of generating interactive game content for developers and researchers, representing an incremental advancement by applying existing LLM capabilities to a new domain without fine-tuning.
The paper tackles the challenge of using large language models (LLMs) for procedural content generation (PCG) by introducing Word2World, a system that enables LLMs to design playable games through stories without task-specific fine-tuning, resulting in coherent worlds and playable games as validated through testing and ablation studies.
Large Language Models (LLMs) have proven their worth across a diverse spectrum of disciplines. LLMs have shown great potential in Procedural Content Generation (PCG) as well, but directly generating a level through a pre-trained LLM is still challenging. This work introduces Word2World, a system that enables LLMs to procedurally design playable games through stories, without any task-specific fine-tuning. Word2World leverages the abilities of LLMs to create diverse content and extract information. Combining these abilities, LLMs can create a story for the game, design narrative, and place tiles in appropriate places to create coherent worlds and playable games. We test Word2World with different LLMs and perform a thorough ablation study to validate each step. We open-source the code at https://github.com/umair-nasir14/Word2World.