PTCBENCH: Benchmarking Contextual Stability of Personality Traits in LLM Systems

TL;DR

This paper introduces PTCBENCH, a benchmark for evaluating how consistent large language model personalities remain across various contexts, highlighting significant personality shifts triggered by external scenarios.

Contribution

It presents a systematic framework for measuring LLM personality stability under diverse situational conditions, addressing a gap in existing research.

Findings

Certain external scenarios cause significant personality changes in LLMs

Personality shifts can impact LLM reasoning capabilities

PTCBENCH provides an extensible framework for realistic environment evaluation

Abstract

With the increasing deployment of large language models (LLMs) in affective agents and AI systems, maintaining a consistent and authentic LLM personality becomes critical for user trust and engagement. However, existing work overlooks a fundamental psychological consensus that personality traits are dynamic and context-dependent. To bridge this gap, we introduce PTCBENCH, a systematic benchmark designed to quantify the consistency of LLM personalities under controlled situational contexts. PTCBENCH subjects models to 12 distinct external conditions spanning diverse location contexts and life events, and rigorously assesses the personality using the NEO Five-Factor Inventory. Our study on 39,240 personality trait records reveals that certain external scenarios (e.g., "Unemployment") can trigger significant personality changes of LLMs, and even alter their reasoning capabilities. Overall, PTCBENCH establishes an extensible framework for evaluating personality consistency in realistic, evolving environments, offering actionable insights for developing robust and psychologically aligned AI systems.