Distributive Fairness in Large Language Models: Evaluating Alignment with Human Values

2025-02-01

AI Summary

This study systematically evaluates large language models’ (LLMs) capacity to model distributive justice principles—such as equality, envy-freeness, and Rawlsian maximin—and their alignment with human fairness preferences in resource allocation. We propose a multidimensional fairness-aware evaluation framework integrating controlled prompt engineering, a menu-based choice paradigm, and a human-annotated dataset for cross-model benchmarking. Our findings reveal that generative allocations produced by LLMs significantly deviate from human fairness intuitions; introducing menu-based choices improves fairness judgment accuracy by up to 47% for certain models; semantic and non-semantic prompts exhibit notable fragility in fairness reasoning; and monetary transfers fail to effectively mitigate inequality under current LLM capabilities. Based on these insights, we propose prompt design guidelines and fine-tuning strategies to enhance fairness alignment, offering both theoretical foundations and practical pathways toward building fairness-aware AI systems.

Technology Category

• Game Theory and Economic Paradigms: Fair Division
• Philosophy and Ethics of AI: Bias, Fairness & Equity
• Humans and AI: Learning Human Values and Preferences

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

The growing interest in employing large language models (LLMs) for decision-making in social and economic contexts has raised questions about their potential to function as agents in these domains. A significant number of societal problems involve the distribution of resources, where fairness, along with economic efficiency, play a critical role in the desirability of outcomes. In this paper, we examine whether LLM responses adhere to fundamental fairness concepts such as equitability, envy-freeness, and Rawlsian maximin, and investigate their alignment with human preferences. We evaluate the performance of several LLMs, providing a comparative benchmark of their ability to reflect these measures. Our results demonstrate a lack of alignment between current LLM responses and human distributional preferences. Moreover, LLMs are unable to utilize money as a transferable resource to mitigate inequality. Nonetheless, we demonstrate a stark contrast when (some) LLMs are tasked with selecting from a predefined menu of options rather than generating one. In addition, we analyze the robustness of LLM responses to variations in semantic factors (e.g. intentions or personas) or non-semantic prompting changes (e.g. templates or orderings). Finally, we highlight potential strategies aimed at enhancing the alignment of LLM behavior with well-established fairness concepts.

Problem

• Large Language Models
• Social Equity
• Fairness Understanding

Innovation

• Large Language Models
• Social Equity Evaluation
• Enhanced Fairness in Decision-making