Creative Generation via Multi-Agent Debate: Does Debate Suppress Diversity?

2026-09-01

AI Summary

研究探讨了多智能体辩论在创意生成任务中抑制多样性的问题，提出Creative-MAD方法，通过认知镜头分配和基于嵌入的同伴选择来维持智能体差异，提高输出多样性。

Abstract

Creative generation tasks, such as narrative writing and scientific ideation, demand both high-quality outputs and distinct responses across independent runs to maximize exploration. Multi-Agent Debate (MAD) has shown strong quality gains on factual and reasoning tasks, making it a natural candidate for creative generation. However, we find its convergence-driven design actively suppresses output diversity across independent runs, creating an inherent trade-off with creative tasks. We theoretically show that preserving diversity among agents within each debate session is a necessary condition for achieving diverse outputs across independent runs. Building on this finding, we propose Creative-MAD, which introduces two synergistic interventions to sustain agent divergence. Specifically, Cognitive Lens Assignment counters identity drift by anchoring each agent to a distinct and persistent cognitive mode, while Embedding-based Peer Selection counters majority pull by limiting each agent's context to its most semantically distant peers. Experiments across four creative benchmarks demonstrate that Creative-MAD significantly enhances both lexical and semantic diversity while maintaining MAD's output quality.

Problem

• Creative Generation
• Multi-Agent Debate
• Diversity Suppression
• Exploration

Innovation

• Creative-MAD
• Cognitive Lens Assignment
• Embedding-based Peer Selection