Ontologies in Design: How Imagining a Tree Reveals Possibilites and Assumptions in Large Language Models

2025-04-03

AI Summary

Current LLM ethics research overemphasizes values (e.g., bias) while neglecting implicit ontological assumptions—i.e., what entities, relations, and possibilities the system presupposes as expressible. Method: The authors introduce the first design-oriented ontological analysis framework, distilling four ontological orientations—plurality, embodiment, vitality, and praxis—and operationalizing ontology as a practical, end-to-end design guide across the LLM development lifecycle. They apply prompt engineering analysis, deconstruction of LLM agent architectures, and socio-technical qualitative research to examine ontological constraints in mainstream chatbots (e.g., reductive “tree” conceptualizations) and static, decontextualized ontologies in agent simulations. Contribution/Results: The work delivers reusable, ontology-sensitive design methods, advancing generative AI toward richer, contextually embedded, and relationally grounded cognitive paradigms.

Technology Category

• Knowledge Representation and Reasoning: Ontologies
• Philosophy and Ethics of AI: Morality & Value-based AI
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Amid the recent uptake of Generative AI, sociotechnical scholars and critics have traced a multitude of resulting harms, with analyses largely focused on values and axiology (e.g., bias). While value-based analyses are crucial, we argue that ontologies -- concerning what we allow ourselves to think or talk about -- is a vital but under-recognized dimension in analyzing these systems. Proposing a need for a practice-based engagement with ontologies, we offer four orientations for considering ontologies in design: pluralism, groundedness, liveliness, and enactment. We share examples of potentialities that are opened up through these orientations across the entire LLM development pipeline by conducting two ontological analyses: examining the responses of four LLM-based chatbots in a prompting exercise, and analyzing the architecture of an LLM-based agent simulation. We conclude by sharing opportunities and limitations of working with ontologies in the design and development of sociotechnical systems.

Problem

• Analyzing ontological assumptions in Large Language Models
• Exploring design orientations for pluralistic AI ontologies
• Assessing LLM limitations through chatbot response analysis

Innovation

• Ontological analysis in LLM design
• Four orientations for ontology consideration
• Examining chatbot responses and simulations