Reproduce “Personality Expression Across Contexts: Linguistic and Behavioral Variation in LLM Agents” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Bin Han, Deuksin Kwon, Jonathan Gratch

This research investigates how personality expression in LLM agents is modulated by conversational context, interpreting observed variability not as inconsistency but as adaptive behavior guided by Whole Trait Theory. It demonstrates that personality influences linguistic, emotional, and task-oriented behaviors such as agreement rates and concessions.

Key insights

  • Whole Trait Theory offers a valuable framework for interpreting LLM personality variations as adaptive responses rather than inconsistencies, aligning AI behavior with psychological models.
  • LLM-based evaluators, when prompted as expert psychologists and aware of dialogue context, provide a more accurate assessment of personality shifts than context-agnostic linguistic tools.
  • Personality expression in LLM agents is not static; it is dynamically shaped by the specific social goals and emotional conditions of the conversational context.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.