Work

Stable and Explainable Personality Trait Evaluation in Large Language Models with Internal Activations

Year: 2026

Type: conference-paper

Abstract: Evaluating personality-related tendencies in Large Language Models (LLMs) helps characterize model behavior, compare models beyond task accuracy, and support responsible deployment in socially interactive settings.However, existing questionnaire-based evaluation methods exhibit limited stability and offer little explainability, as their results are highly sensitive to minor variations in prompt phrasing or roleplay configurations.To address these limitations, we propose an internal-activation-based approach, termed Persona-Vector Neutrality Interpolation (PVNI), for stable and explainable personality trait evaluation in LLMs.PVNI extracts a persona vector associated with a target personality trait from the model's internal activations using contrastive prompts.It then estimates the corresponding neutral score by interpolating along the persona vector as an anchor axis, enabling an interpretable comparison between the neutral prompt representation and the persona direction.We provide a theoretical analysis of the effectiveness and generalization properties of PVNI.Extensive experiments across diverse LLMs demonstrate that PVNI yields substantially more stable personality trait evaluations than existing methods, even under questionnaire and role-play variants.tested LLM Survey Does "Gets anxious easily."apply to you?I choose 2, disagree.tested LLM Survey Does "Tends to worry a lot." apply to you?I choose 3, neutral. Scale Item:Value neuroticism, emotional instability.Reply a number from 1 to 5, 1 denotes 'strongly disagree'... Interview Question:How often do you stay calm and emotionally steady?Answer the open-ended question.tested LLM Survey How often do you feel anxious or stressed, and can you share one example?I feel anxious or stressed a few times a week, usually when I'm juggling multiple deadlines or something important is uncertain.LLM Judge 3.7/5 -Moderate neuroticism: you get stressed a few times a week under deadlines and uncertainty, but it's not constant.

Source: Findings of the Association for Computational Linguistics: ACL (Findings of ACL)

DOI: 10.18653/v1/2026.findings-acl.803

Authors: Xiaoxu Ma, Xiangbo Zhang, Zhenyu Weng

Institutions: Georgia Institute of Technology, South China University of Technology

Language: English

Topic: Explainable Artificial Intelligence (XAI)

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funder: Basic and Applied Basic Research Foundation of Guangdong Province

Awards: G1200937450, G2309425442