Homogenization in Multi-Agent Systems

Key points

• Multi-agent systems (MAS) leverage interactions between agents to perform complex tasks.1
• Despite their success, we show that these interactions can also lead to homogenization, i.e., agents converging to similar behaviors.1
• Across these tasks, we show that homogenization translates to concrete downstream risks: in code generation, it hides and amplifies correlated errors which can create systemic vulnerabilities; in hiring, it allows the influence of biased agents to persist long after their removal; and in peer review, it creates uneven evaluation standards across research areas.1
• Finally, we show that simple approaches to increase diversity---leveraging sampling stochasticity and mixed-models MAS---fail to reduce homogenization risks, highlighting the need for strategies to effectively leverage agent diversity.1

Sources (1)

• [1]Homogenization in Multi-Agent Systems
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 10:49 AM

  Homogenization in MAS can reduce agent diversity and reinforce shared failures.

  Multi-agent systems (MAS) leverage interactions between agents to perform complex tasks.

Extractive summary: sentences quoted from the sources.