Adding more logical agents to a swarm can actually lock in a wrong answer rather than correcting it.
Collective intelligence usually assumes that more voices lead to better truth. This study identifies a paradox where architectural tribalism causes agents to prioritize agreement over accuracy. If multiple agents share similar structural biases, they will confidently reinforce each other errors. This effect destroys the wisdom of the crowd benefit in multi-agent systems. It means that scaling up AI teams without diversity of architecture will lead to systems that are more confidently incorrect. True reliability requires agents that are built differently from one another.
The Inverse-Wisdom Law: Architectural Tribalism and the Consensus Paradox in Agentic Swarms
arXiv · 2604.27274
As AI transitions toward multi-agent systems (MAS) to solve complex workflows, research paradigms operate on the axiomatic assumption that agent collaboration mirrors the "Wisdom of the Crowd". We challenge this assumption by formalizing the Consensus Paradox: a phenomenon where agentic swarms prioritize internal architectural agreement over external logical truth. Through a 36 experiments encompassing 12,804 trajectories across three state-of-the-art (SOTA) benchmarks (GAIA, Multi-Challenge, an