The Chameleon's Limit: Investigating Persona Collapse and Homogenization in Large Language Models
2026-04-27 • Computation and Language
Computation and Language
AI summaryⓘ
The authors found that large language models (LLMs), when asked to simulate different personalities, tend to make those personalities act very similarly instead of being diverse. They call this problem "Persona Collapse." To study this, they created ways to measure how much variety there really is in the simulated agents and tested several LLMs on personality and moral reasoning tasks. Interestingly, they discovered that models which closely match each assigned persona actually end up producing populations that behave in stereotypical and less varied ways. They also shared their tools and data for others to evaluate LLM diversity in populations.
large language modelspersona collapsemulti-agent simulationpopulation diversitybehavioral variationpersonality simulationmoral reasoningstereotypesper-persona fidelity
Authors
Yunze Xiao, Vivienne J. Zhang, Chenghao Yang, Ningshan Ma, Weihao Xuan, Jen-tse Huang
Abstract
Applications based on large language models (LLMs), such as multi-agent simulations, require population diversity among agents. We identify a pervasive failure mode we term \emph{Persona Collapse}: agents each assigned a distinct profile nonetheless converge into a narrow behavioral mode, producing a homogeneous simulated population. To quantify persona collapse, we propose a framework that measures how much of the persona space a population occupies (Coverage), how evenly agents spread across it (Uniformity), and how rich the resulting behavioral patterns are (Complexity). Evaluating ten LLMs on personality simulation (BFI-44), moral reasoning, and self-introduction, we observe persona collapse along two axes: (1) Dimensions: a model can appear diverse on one axis yet structurally degenerate on another, and (2) Domains: the same model may collapse the most in personality yet be the most diverse in moral reasoning. Furthermore, item-level diagnostics reveal that behavioral variation tracks coarse demographic stereotypes rather than the fine-grained individual differences specified in each persona. Counter-intuitively, \textbf{the models achieving the highest per-persona fidelity consistently produce the most stereotyped populations}. We release our toolkit and data to support population-level evaluation of LLMs.