Ambiguity Collapse by LLMs: A Taxonomy of Epistemic Risks

TL;DR

This paper introduces the concept of ambiguity collapse in large language models, analyzing its epistemic risks at multiple levels and proposing mitigation strategies to preserve meaningful ambiguity in AI systems.

Contribution

It develops a taxonomy of epistemic risks caused by ambiguity collapse in LLMs and offers multi-layer mitigation principles to address these challenges.

Findings

Identifies three levels of epistemic risks: process, output, ecosystem.

Provides case studies illustrating the risks of ambiguity collapse.

Proposes mitigation strategies across training, deployment, and interface design.

Abstract

Large language models (LLMs) are increasingly used to make sense of ambiguous, open-textured, value-laden terms. Platforms routinely rely on LLMs for content moderation, asking them to label text based on disputed concepts like "hate speech" or "incitement"; hiring managers may use LLMs to rank who counts as "qualified"; and AI labs increasingly train models to self-regulate under constitutional-style ambiguous principles such as "biased" or "legitimate". This paper introduces ambiguity collapse: a phenomenon that occurs when an LLM encounters a term that genuinely admits multiple legitimate interpretations, yet produces a singular resolution, in ways that bypass the human practices through which meaning is ordinarily negotiated, contested, and justified. Drawing on interdisciplinary accounts of ambiguity as a productive epistemic resource, we develop a taxonomy of the epistemic risks posed by ambiguity collapse at three levels: process (foreclosing opportunities to deliberate, develop cognitive skills, and shape contested terms), output (distorting the concepts and reasons agents act upon), and ecosystem (reshaping shared vocabularies, interpretive norms, and how concepts evolve over time). We illustrate these risks through three case studies, and conclude by sketching multi-layer mitigation principles spanning training, institutional deployment design, interface affordances, and the management of underspecified prompts, with the goal of designing systems that surface, preserve, and responsibly govern ambiguity.