Multi-agent debates show bias toward first speaker but personality prompts help
When Order Matters: First-Speaker Bias and Mitigation through Personality in Sequential Multi-Agent Debate
Artificial Intelligence
Summary
When multiple AI agents take turns debating, the first speaker often has a big advantage in shaping the final decision. The researchers found that putting a stronger AI agent last can reduce its influence. They tested ways to change how agents behave by adjusting personality traits like agreeableness and extraversion. Making the strong agent less agreeable helped restore its influence and improved the overall accuracy. This shows that who speaks first and the agents’ interaction styles both impact how debates unfold.
What this means in practice
- •For ai system developers: Improve decision-making accuracy in AI systems by designing agent interaction protocols that control speaking order and induce specific personality traits.
- •For conversational ai builders: Create more balanced multi-agent chatbot dialogues by using personality prompting to reduce bias from dialogue order.
Authors
Duofeng Xu, Bryan Hooi, Dandan Qiao
Abstract
Multi-agent debate (MAD) is often used to improve large language model (LLM) reasoning, but sequential debate is rarely a neutral aggregator of agents' opinions. We show that sequential MAD suffers from a pronounced first-speaker bias: agents disproportionately shape the final answer when they speak first. As a result, placing a stronger model after weaker ones can substantially offset its reasoning advantage. We then focus on the disadvantaged strong-agent-last setting and ask whether personality prompting can mitigate this imbalance. Drawing on the Big Five model, we study agreeableness and extraversion as behavioral interventions applied to either the strong or weak side. We find that their effects are trait-specific. Influence consistently shifts in the direction of lower agreeableness, and assigning low agreeableness to the stronger agent helps restore its lost influence and improves final accuracy. Extraversion, by contrast, produces less systematic changes in influence and accuracy, with its clearest effect appearing in agents' verbosity. These findings show that effective MAD design depends not only on model capability, but also on how speaking order and induced interaction behavior shape the debate process.