Can Large Language Models be Good Emotional Supporter? Mitigating Preference Bias on Emotional Support Conversation

2024-02-20

Annual Meeting of the Association for Computational Linguistics

AI Summary

Large language models (LLMs) exhibit strategy preference bias in emotional support conversations (ESCs), degrading response appropriateness and robustness. Method: This study is the first to systematically demonstrate a significant negative correlation between LLMs’ strategy preference intensity and emotional support efficacy, empirically validated on the ESConv dataset via strategy distribution analysis, quantitative preference assessment, and comparative experiments involving prompt engineering and fine-tuning. Key findings include: both extremely low and high preference intensities impair support quality; external strategy guidance effectively mitigates bias; and pure LLM architectures fail to match human-level support performance. Contribution/Results: We identify strategy preference bias as a critical failure mechanism in ESCs, propose an intervenable evaluation framework for quantifying and addressing such bias, and provide theoretical foundations and practical pathways for developing human-AI collaborative emotional support systems.

Technology Category

• Humans and AI: Emotional Intelligence
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

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
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions

Abstract

Emotional Support Conversation (ESC) is a task aimed at alleviating individuals' emotional distress through daily conversation. Given its inherent complexity and non-intuitive nature, ESConv dataset incorporates support strategies to facilitate the generation of appropriate responses. Recently, despite the remarkable conversational ability of large language models (LLMs), previous studies have suggested that they often struggle with providing useful emotional support. Hence, this work initially analyzes the results of LLMs on ESConv, revealing challenges in selecting the correct strategy and a notable preference for a specific strategy. Motivated by these, we explore the impact of the inherent preference in LLMs on providing emotional support, and consequently, we observe that exhibiting high preference for specific strategies hinders effective emotional support, aggravating its robustness in predicting the appropriate strategy. Moreover, we conduct a methodological study to offer insights into the necessary approaches for LLMs to serve as proficient emotional supporters. Our findings emphasize that (1) low preference for specific strategies hinders the progress of emotional support, (2) external assistance helps reduce preference bias, and (3) existing LLMs alone cannot become good emotional supporters. These insights suggest promising avenues for future research to enhance the emotional intelligence of LLMs.

Problem

• LLMs struggle with selecting correct emotional support strategies
• High preference for specific strategies hinders effective emotional support
• Existing LLMs alone cannot become proficient emotional supporters

Innovation

• Analyzing LLMs' strategy selection challenges in ESC
• Exploring impact of LLMs' preference on emotional support
• Proposing external assistance to reduce preference bias