Author Identifier (ORCID)
Abstract
Large Language Models (LLMs) are widely used as conversational agents, exploiting their capabilities in various sectors such as edu- cation, law, medicine, and more. However, LLMs are often subjected to context-shifting behaviour, resulting in a lack of consistent and interpretable personality-aligned interactions. Adherence to psychological traits lacks com- prehensive analysis, especially in the case of dyadic (pairwise) conversations. We examine this challenge from two viewpoints, initially using two conversation agents to generate a dis- course on a certain topic with an assigned per- sonality from the OCEAN framework (Open- ness, Conscientiousness, Extraversion, Agree- ableness, and Neuroticism) as High/Low for each trait. This is followed by using multi- ple judge agents to infer the original traits as- signed to explore prediction consistency, inter- model agreement, and alignment with the as- signed personality. Our findings indicate that while LLMs can be guided toward personality- driven dialogue, their ability to maintain per- sonality traits varies significantly depending on the combination of models and discourse settings. These inconsistencies emphasise the challenges in achieving stable and interpretable personality-aligned interactions in LLMs
Keywords
large language models, LLM's
Document Type
Conference Proceeding
Date of Publication
2025
Publication Title
Conference on Empirical Methods in Natural Language Processing
Publisher
Association for Computational Linguistics
School
School of Business and Law
RAS ID
84234
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
ISBN
979-8-89176-332-6
Recommended Citation
Bhandari, P., Fay, N., Wise, M. J., Datta, A., Meek, S., Naseem, U., & Nasim, M. (2025). Can LLM agents maintain a persona in discourse? In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing (pp. 29201–29217).https://doi.org/10.18653/v1/2025.emnlp-main.1487