Prompt Optimization Across Multiple Agents for Representing Diverse Human Populations

2025-10-08

AI Summary

Large language models (LLMs) often produce homogeneous outputs when simulating human behavior, failing to capture population-level diversity. Method: This paper proposes a submodular optimization-based framework for constructing diverse multi-agent systems. Leveraging a small set of human demonstration data, it employs in-context learning and prompt engineering to elicit heterogeneous responses from LLMs, then efficiently selects the most representative subset of agents from an exponential-sized candidate space. Contribution/Results: The core innovation lies in formalizing behavioral diversity as a submodular function maximization problem, enabling polynomial-time algorithms with theoretical approximation guarantees. Experiments across crowdsourcing and educational settings demonstrate that our approach significantly outperforms single-agent baselines and existing methods, achieving higher fidelity in both behavioral pattern reproduction and opinion distribution alignment.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Search and Optimization: Learning to Search
• Multiagent Systems: Multiagent Learning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

The difficulty and expense of obtaining large-scale human responses make Large Language Models (LLMs) an attractive alternative and a promising proxy for human behavior. However, prior work shows that LLMs often produce homogeneous outputs that fail to capture the rich diversity of human perspectives and behaviors. Thus, rather than trying to capture this diversity with a single LLM agent, we propose a novel framework to construct a set of agents that collectively capture the diversity of a given human population. Each agent is an LLM whose behavior is steered by conditioning on a small set of human demonstrations (task-response pairs) through in-context learning. The central challenge is therefore to select a representative set of LLM agents from the exponentially large space of possible agents. We tackle this selection problem from the lens of submodular optimization. In particular, we develop methods that offer different trade-offs regarding time complexity and performance guarantees. Extensive experiments in crowdsourcing and educational domains demonstrate that our approach constructs agents that more effectively represent human populations compared to baselines. Moreover, behavioral analyses on new tasks show that these agents reproduce the behavior patterns and perspectives of the students and annotators they are designed to represent.

Problem

• Optimizing prompt selection for diverse agent representation
• Addressing homogeneous LLM outputs lacking human diversity
• Constructing representative agent sets via submodular optimization

Innovation

• Multiple LLM agents represent human diversity
• Agents conditioned on human demonstrations via in-context learning
• Submodular optimization selects representative agent sets