通过四阶段提示提升大模型共情能力，减少社会偏见。
Empathetic Cascading Networks: A Multi-Stage Prompting Technique for Reducing Social Biases in Large Language Models
- 分四步引导模型理解他人情感与处境。
- 在GPT-3.5-turbo和GPT-4上获得最高共情得分。
- 适合需要高共情与包容性的对话AI场景。
本报告提出情感级联网络（ECN）框架，一种多阶段提示方法，旨在增强大语言模型的共情与包容能力。ECN包含四个阶段：视角采纳、情感共鸣、反思理解与整合合成，引导模型生成具有情感共鸣且情境敏感的回应。实验结果表明，ECN在GPT-3.5-turbo和GPT-4上均取得最高共情指数（EQ）得分，同时保持良好的尊重度（Regard）与困惑度（Perplexity）表现。这些发现凸显了该方法在需高度共情与包容性的对话AI应用中的潜力。
原文摘要 · Abstract (English)
This report presents the Empathetic Cascading Networks (ECN) framework, a multi-stage prompting method designed to enhance the empathetic and inclusive capabilities of large language models. ECN employs four stages: Perspective Adoption, Emotional Resonance, Reflective Understanding, and Integrative Synthesis, to guide models toward generating emotionally resonant and contextually aware responses. Experimental results demonstrate that ECN achieves the highest Empathy Quotient (EQ) scores across GPT-3.5-turbo and GPT-4, while maintaining competitive Regard and Perplexity metrics. These findings emphasize ECN's potential for applications requiring empathy and inclusivity in conversational AI.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。