From Personas to Talks: Revisiting the Impact of Personas on LLM-Synthesized Emotional Support Conversations

2025-02-17

AI Summary

This study investigates how Big Five personality traits influence emotional support dialogue generation in large language models (LLMs). To address the problem of uncontrolled and inconsistent persona expression, we inject quantifiable personality dimensions via prompt engineering and systematically analyze their effects using ESTA-based annotation, multi-dimensional human evaluation, and an automated Persona-Consistency Score. Our method reveals, for the first time, implicit personality drift patterns in LLMs during emotional support interactions. Empirical results demonstrate that fine-tuning for neuroticism and extraversion significantly improves empathic response relevance (+23.6%) and personalized strategy coverage (+31.4%), while also enhancing dialogue pacing and strategy adaptability. We establish a “personality–strategy–outcome” association model, offering both theoretical foundations and technical pathways for controllable, trustworthy emotional support dialogue generation.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Humans and AI: Emotional Intelligence

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

The rapid advancement of Large Language Models (LLMs) has revolutionized the generation of emotional support conversations (ESC), offering scalable solutions with reduced costs and enhanced data privacy. This paper explores the role of personas in the creation of ESC by LLMs. Our research utilizes established psychological frameworks to measure and infuse persona traits into LLMs, which then generate dialogues in the emotional support scenario. We conduct extensive evaluations to understand the stability of persona traits in dialogues, examining shifts in traits post-generation and their impact on dialogue quality and strategy distribution. Experimental results reveal several notable findings: 1) LLMs can infer core persona traits, 2) subtle shifts in emotionality and extraversion occur, influencing the dialogue dynamics, and 3) the application of persona traits modifies the distribution of emotional support strategies, enhancing the relevance and empathetic quality of the responses. These findings highlight the potential of persona-driven LLMs in crafting more personalized, empathetic, and effective emotional support dialogues, which has significant implications for the future design of AI-driven emotional support systems.

Problem

• Role of personas in LLM-generated emotional support
• Stability of persona traits in dialogues
• Impact of personas on dialogue quality and strategy

Innovation

• LLMs infuse persona traits
• Psychological frameworks measure traits
• Persona traits modify dialogue strategy