Modeling Motivated Reasoning in Law: Evaluating Strategic Role Conditioning in LLM Summarization

2025-08-30

AI Summary

This study investigates whether large language models (LLMs) exhibit motivated reasoning—systematic, role-congruent bias in factual selection and argument construction—when generating role-specific legal summaries (e.g., for judges, prosecutors, or attorneys). Method: Employing a role-conditioned prompting design grounded in legal realism, we develop a dual-dimensional evaluation framework measuring both legal fact coverage and stakeholder-oriented bias. Experiments explicitly test whether “neutrality” instructions mitigate role-driven distortions. Contribution/Results: Even under neutrality constraints, LLMs exhibit statistically significant, role-consistent information filtering—implicitly inferring and reinforcing position-specific stances. This work provides the first quantitative evidence of structural stance bias in legal summarization by LLMs. It introduces the “role-aware evaluation” paradigm, establishing a methodological foundation for identifying and governing alignment risks in high-stakes legal AI applications.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Computer Vision: Bias, Fairness & Privacy

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Large Language Models (LLMs) are increasingly used to generate user-tailored summaries, adapting outputs to specific stakeholders. In legal contexts, this raises important questions about motivated reasoning -- how models strategically frame information to align with a stakeholder's position within the legal system. Building on theories of legal realism and recent trends in legal practice, we investigate how LLMs respond to prompts conditioned on different legal roles (e.g., judges, prosecutors, attorneys) when summarizing judicial decisions. We introduce an evaluation framework grounded in legal fact and reasoning inclusion, also considering favorability towards stakeholders. Our results show that even when prompts include balancing instructions, models exhibit selective inclusion patterns that reflect role-consistent perspectives. These findings raise broader concerns about how similar alignment may emerge as LLMs begin to infer user roles from prior interactions or context, even without explicit role instructions. Our results underscore the need for role-aware evaluation of LLM summarization behavior in high-stakes legal settings.

Problem

• Evaluating strategic role conditioning in LLM legal summarization
• Investigating how LLMs frame information for different legal stakeholders
• Assessing selective inclusion patterns in role-conditioned judicial summaries

Innovation

• Role-conditioned LLM summarization for legal stakeholders
• Evaluation framework for legal fact and reasoning inclusion
• Analysis of selective pattern alignment with legal roles