AI agents reshape consensus formation in human groups
Lin Chen, Ziyi Liu, Xia Hu and Yong Li run a collaborative description game with mixed human and LLM-agent groups and find three distinct regimes of consensus formation as the agent proportion rises.
Ask this paper
Three regimes, and the middle one is the bad one: Low agent proportions facilitate human-led consensus, intermediate proportions disrupt convergence entirely, and high proportions restore strong consensus but shift it to agent-led conventions.
The content of consensus changes, not just its strength: Human-led consensus is concrete, holistic and grounded in shared real-world analogies; agent-led consensus is more abstract, less information-dense and more geometrically segmented.
A mechanistic account: Agent influence comes from a shared linguistic prior that places agents near one another in expression space, combined with unusually stable expression choices across rounds.
Humans resist, then conform: Participants initially decline to adopt expressions from partners they perceive as AI, then gradually yield to conformity pressure.
Why it matters: Agent proportion and agent transparency become design variables with measurable effects on group norms, which is a concrete handle for anyone deploying agents into human teams or forums.
Abstract
As large language model (LLM) agents shift from tools to participants in human groups, a fundamental question for collective behavior is how their growing presence reshapes consensus formation. Here we study mixed human-AI groups in a collaborative description game, in which shared conventions emerge through repeated rounds of random pairwise communication. Varying the proportions of LLM agents, we identify three distinct regimes of consensus formation: low agent proportions facilitate human-led consensus, intermediate proportions disrupt convergence, and high proportions restore strong consensus while shifting it toward agent-led conventions. Crucially, these regimes differ not only in the strength of convergence, but also in the semantic grounding and communicative form of the resulting consensus: human-led consensus is more concrete, holistic, and grounded in shared real-world analogies, whereas agent-led consensus is more abstract, less information-dense, and more geometrically segmented. Mechanistically, agent influence arises from a shared linguistic prior that places agents near one another in the expression space, combined with relatively stable expression choices across rounds; humans initially resist adopting expressions from partners perceived as AI but gradually yield to conformity pressure. These findings provide evidence that AI composition can shape the emergence, content, and perceived legitimacy of group norms, making agent proportion and transparency important design variables for human-AI systems.