Multi-agent systems converge to the strongest model in consensus
Who Is the Influencer? Consensus Dynamics in Multi-Agent Systems
Artificial Intelligence
Summary
When several AI models work together to reach an agreement, the model with more parameters tends to have the most influence on the final outcome. This happens even if the task doesn’t require any prior knowledge and has no right or wrong answer. The authors explain this by comparing models to weighted particles, where bigger models have more weight and pull the group’s opinion toward their own. This discovery helps clarify how groups of AI models interact and why larger models often dominate decisions.
What this means in practice
- •For ai system developers: Design multi-agent AI systems that predict which model will guide the group decision based on model size.
- •For collaborative robotics teams: Use understanding of influence dynamics to balance contributions from robots with different processing strengths in team tasks.
Authors
Gouki Minegishi, Jingyuan Feng, Hiroki Furuta, Takeshi Kojima, Yusuke Iwasawa, Yutaka Matsuo
Abstract
Multi-agent LLM systems are increasingly used in practice, but their behavior is complex and hard to understand. We design a simple task to ask which model has the most influence on the final consensus when several models are mixed in one system. Experimentally, across seven model families, we find that the opinions of the other models are absorbed into the opinion of the stronger model, the one with more parameters. This happens even though the task requires no pretraining knowledge and has no correct answer. To understand this phenomenon, we interpret the agents as weighted particles moving on a simplex. Larger models carry a larger weight and therefore occupy a deeper well in the energy landscape. A large weight means strong in-context learning: a model with a large weight amplifies the majority of its context. Larger models are generally better at in-context learning, and this is what makes the larger model the influencer. We hope that this work helps to interpret multi-agent systems, whose behavior is becoming too complex for humans to understand.