Single-Agent Generation Surpasses Multi-Agent Systems in Semantic Diversity

ACL ARR 2026 January Submission4722 Authors

Keywords: Multi-agent system, Diversity, Single-Agent vs. Multi-Agent, Open-Ended question, Divergent thinking, Multi-Output Generation

Abstract:

Multi-Agent Systems (MAS) are commonly used to improve reasoning diversity and robustness by simulating interactions among agents with distinct roles. However, prior work often entangles the contribution of the multi-agent architecture with that of prompt conditioning, making the source of observed diversity gains unclear.
We address this confound with a controlled study on divergent thinking tasks, using identical prompt conditioning for MAS and single-agent settings. Under these matched conditions, single-agent setups consistently outperform multi-agent systems in semantic diversity. We attribute this gap to \textit{information visibility}: parallel agents often converge on overlapping ideas, whereas a serial single-agent model can condition on its own generation history to avoid redundancy.
We further find that a Multi-Output strategy, which prompts a single agent to produce multiple responses within a single inference pass, achieves the highest diversity without degrading logical validity. Together, these results suggest that costly inter-agent interactions may be unnecessary; instead, efficient information sharing and multi-output generation appear to be effective mechanisms for improving diversity, with implications for designing efficient agentic frameworks.

Paper Type: Long

Research Area: AI/LLM Agents

Research Area Keywords: Language Modeling, Interpretability and Analysis of Models for NLP, Generation

Contribution Types: NLP engineering experiment

Languages Studied: English

Submission Number: 4722