Critical Confabulation: Can LLMs Hallucinate for Social Good?

2025-11-11

AI Summary

Historical archives systematically omit marginalized individuals due to structural injustice, resulting in narrative erasure and the absence of “hidden figures.” Method: This paper proposes a *critical fabulation* framework that harnesses controlled hallucination in large language models (LLMs) as a socially just knowledge-production mechanism. Leveraging the open-source OLMo-2 model, we integrate narrative completion tasks, unpublished historical corpora, and multi-strategy prompt engineering—strictly constrained by evidentiary grounding—to reconstruct diverse, credible biographical narratives of marginalized subjects. Contribution/Results: Empirical evaluation demonstrates that LLMs can perform *evidence-anchored inference*: generated narratives remain faithful to extant archival records while plausibly filling structural gaps. This work constitutes the first theoretical formalization of controlled hallucination as a historiographical methodology, thereby expanding the boundaries of responsible LLM application in the humanities and social sciences.

Technology Category

• Humans and AI: Game Design — Procedural Content Generation & Storytelling
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Natural Language Processing: (Large) Language Models

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

LLMs hallucinate, yet some confabulations can have social affordances if carefully bounded. We propose critical confabulation (inspired by critical fabulation from literary and social theory), the use of LLM hallucinations to"fill-in-the-gap"for omissions in archives due to social and political inequality, and reconstruct divergent yet evidence-bound narratives for history's"hidden figures". We simulate these gaps with an open-ended narrative cloze task: asking LLMs to generate a masked event in a character-centric timeline sourced from a novel corpus of unpublished texts. We evaluate audited (for data contamination), fully-open models (the OLMo-2 family) and unaudited open-weight and proprietary baselines under a range of prompts designed to elicit controlled and useful hallucinations. Our findings validate LLMs'foundational narrative understanding capabilities to perform critical confabulation, and show how controlled and well-specified hallucinations can support LLM applications for knowledge production without collapsing speculation into a lack of historical accuracy and fidelity.

Problem

• Using LLM hallucinations to fill gaps in archives from social inequality
• Reconstructing evidence-bound narratives for history's hidden figures
• Developing controlled hallucinations for knowledge production without sacrificing accuracy

Innovation

• Using LLM hallucinations to fill archival gaps
• Simulating narrative gaps with cloze tasks
• Controlling hallucinations for knowledge production