Games as Ontology Engines: AI and LLMs Invoke Spatiotemporal and Metaphysical Realities in Virtual Worlds

Jasmine Roberts, Andrzej Banburski

Keywords: AI-driven content generation, AI-driven Gameplay, Virtual Reality, Large Language Models, Ontology, Semantic Networks, Symbolic Reasoning, Human-Computer Interaction

TL;DR: The paper shows how games use AI and real-time object transformation to create dynamic ontologies, turning gameplay into an exploration of metaphysical relationships.

Abstract: Contrary to common perception, games are not solely sources of entertainment; they are powerful mediums for modeling and communicating complex relationships. By design, games are composable systems for representing and manipulating interactions, especially the spatiotemporal. Historically, these interactions have been based around causal chains, which determine the sequencing of player actions and outcomes. In this paper, we repurpose games as ‘ontology engines,’ generating new metaphysical relationships and providing tools for experimenting with them. We explore how virtual worlds supported by large language models (LLMs) and real-time object-transformation systems can serve as frameworks for examining both culturally situated and universally recognizable ontologies.

Submission Number: 25