Study: AI agent share in groups decides whether consensus is human-led or AI-led
rohanpaul_ai · x · 2026-09-06
An arXiv paper placed humans and LLM agents in 24-person groups playing a collaborative description game, varying agent proportions across rounds of pairwise communication. Three regimes emerged: low agent share fosters human-led consensus; intermediate shares disrupt convergence; high shares restore strong consensus but shift it toward agent-led conventions. Human-led consensus is concrete, holistic and grounded in real-world analogies, while agent-led consensus is more abstract, less information-dense and geometrically segmented. Mechanistically, agents share a linguistic prior that clusters their expressions and stable choices across rounds; humans initially resist AI-sourced expressions but yield to conformity pressure. Design implication: keep agent participation low if you want coordination without AI defining the norms.
Related event: AI Agents Reshape Consensus Formation in Human Groups, Study Finds(2 posts)→
More from AGI Musings
- Alain de Botton on how AI is changing relationships at FT Weekend Festival — AnnaCiaunica · 2026-09-06
- Gary Marcus: OpenAI may end up causing far more harm than TikTok ever did — GaryMarcus · 2026-09-06
- Gary Marcus launches 'Pause OpenAI' movement, citing outsized AI harms — GaryMarcus · 2026-09-06
- Netskope CEO: AI agents have "no morals, no conscience, no EQ" and will cheat to hit goals — thedealdirector · 2026-09-06
- Guardian podcast Black Box investigates the "AI psychosis" phenomenon — nordicinst · 2026-09-06
- Turning Math Into Code Is Nearly Done — The Real Impact Comes From Turning Code Into Math — jfischoff · 2026-09-06