arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2404.13071cs.CLcs.AI
收藏

使用大型语言模型对情感与伦理进行建模

Modeling Emotions and Ethics with Large Language Models

Stanford University(斯坦福大学)

Edward Y. Chang

AI总结:

本文提出将类人情感与伦理维度整合入大型语言模型,通过协作式LLMs建模八种对立情感,并引入结合人类反馈的自监督学习算法(SSHF)实现伦理自我评估与调整,推动AI向共情互动与原则性决策发展。

摘要:

本文探讨了将类人情感与伦理考量整合入大型语言模型(LLMs)中。我们首先对八种基本人类情感进行建模,将其作为对立对呈现,并采用协作式LLMs在不同强度范围内重新解释和表达这些情感。我们的研究重点是在LLMs中嵌入潜在的伦理维度,由一种结合人类反馈的新型自监督学习算法(SSHF)进行指导。该方法使LLMs能够针对伦理准则进行自我评估和调整,从而增强其生成既具情感共鸣又符合伦理的内容的能力。本文提出的方法论与案例研究展示了LLMs超越纯文本和图像生成的潜力,进而涉足共情互动与原则性决策的领域,从而在开发具备情感感知与伦理意识的AI系统方面树立了新先例。

英文摘要:

This paper explores the integration of human-like emotions and ethical considerations into Large Language Models (LLMs). We first model eight fundamental human emotions, presented as opposing pairs, and employ collaborative LLMs to reinterpret and express these emotions across a spectrum of intensity. Our focus extends to embedding a latent ethical dimension within LLMs, guided by a novel self-supervised learning algorithm with human feedback (SSHF). This approach enables LLMs to perform self-evaluations and adjustments concerning ethical guidelines, enhancing their capability to generate content that is not only emotionally resonant but also ethically aligned. The methodologies and case studies presented herein illustrate the potential of LLMs to transcend mere text and image generation, venturing into the realms of empathetic interaction and principled decision-making, thereby setting a new precedent in the development of emotionally aware and ethically conscious AI systems.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑