Generative Social Choice: The Next Generation

2025-05-28

AI Summary

This work addresses the problem of generating a length-constrained, proportionally representative list of statements from massive volumes of unstructured user opinions in democratic deliberation settings. To tackle the challenge of an infinite, query-only candidate statement space, we formulate generative social choice as a budget-constrained committee election problem—the first such formalization—and provide theoretically grounded approximation guarantees. Our method integrates social choice theory with large language models (specifically GPT-4o), introducing a queryable generation mechanism and a proportionality-aware optimization algorithm that jointly respects total length budgets and coverage breadth. Experiments on urban governance and pharmaceutical evaluation datasets demonstrate that our generated statement lists significantly improve opinion coverage and reduce group-level bias, validating both effectiveness and practical utility.

Technology Category

• Game Theory and Economic Paradigms: Social Choice / Voting
• Natural Language Processing: Generation
• Knowledge Representation and Reasoning: Preferences

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

A key task in certain democratic processes is to produce a concise slate of statements that proportionally represents the full spectrum of user opinions. This task is similar to committee elections, but unlike traditional settings, the candidate set comprises all possible statements of varying lengths, and so it can only be accessed through specific queries. Combining social choice and large language models, prior work has approached this challenge through a framework of generative social choice. We extend the framework in two fundamental ways, providing theoretical guarantees even in the face of approximately optimal queries and a budget limit on the overall length of the slate. Using GPT-4o to implement queries, we showcase our approach on datasets related to city improvement measures and drug reviews, demonstrating its effectiveness in generating representative slates from unstructured user opinions.

Problem

• Producing concise slates representing diverse user opinions proportionally
• Handling variable-length candidate statements accessible only via queries
• Ensuring theoretical guarantees under query approximation and length constraints

Innovation

• Extends generative social choice framework
• Theoretical guarantees for approximate queries
• Budget limit on slate length