arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2502.02444cs.CLcs.AI
收藏

生成式心理词汇法构建大型语言模型中的价值体系

Generative Psycho-Lexical Approach for Constructing Value Systems in Large Language Models

State Key Laboratory of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University(人工智能通用基础理论国家重点实验室,智能科学与技术学院,北京大学);Yuanpei College, Peking University(元培学院,北京大学);School of Psychological and Cognitive Sciences, Peking University(心理学与认知科学学院,北京大学);Key Laboratory of Machine Perception (Ministry of Education), Peking University(机器感知重点实验室(教育部),北京大学);PKU-Wuhan Institute for Artificial Intelligence(北京大学-武汉人工智能研究院)

Haoran Ye, Tianze Zhang, Yuhang Xie, Liyuan Zhang, Yuanyi Ren, Xin Zhang, Guojie Song

AI总结:

本研究提出生成式心理词汇法(GPLA),构建基于心理学理论的五因素价值体系,以更好地捕捉、预测并增强大型语言模型的价值观对齐与安全性。

摘要:

价值观是个人和集体感知、认知与行为的核心驱动力。价值体系,如施瓦茨的基本人类价值理论,描绘了这些价值观之间的层级与相互作用,从而支持对决策过程和社会动态的跨学科研究。近年来,大型语言模型(LLMs)的兴起引发了对其难以捉摸的内在价值观的担忧。尽管在评估、理解和对齐LLM价值观方面付出了越来越多的努力,但一个基于心理学理论的LLM价值体系仍未得到充分探索。本研究通过引入生成式心理词汇法(GPLA)来弥补这一空白,这是一种可扩展、适应性强且具有理论依据的构建价值体系的方法。利用GPLA,我们提出了一个专为LLM量身定制的、基于心理学理论的五因素价值体系。为了进行系统验证,我们提出了三项基准测试任务,这些任务将心理学原理与前沿的人工智能优先事项相结合。我们的结果表明,与经典的施瓦茨价值观相比,所提出的价值体系符合标准心理学标准,能更好地捕捉LLM的价值观,提高LLM的安全性预测,并增强LLM的对齐性。

英文摘要:

Values are core drivers of individual and collective perception, cognition, and behavior. Value systems, such as Schwartz's Theory of Basic Human Values, delineate the hierarchy and interplay among these values, enabling cross-disciplinary investigations into decision-making and societal dynamics. Recently, the rise of Large Language Models (LLMs) has raised concerns regarding their elusive intrinsic values. Despite growing efforts in evaluating, understanding, and aligning LLM values, a psychologically grounded LLM value system remains underexplored. This study addresses the gap by introducing the Generative Psycho-Lexical Approach (GPLA), a scalable, adaptable, and theoretically informed method for constructing value systems. Leveraging GPLA, we propose a psychologically grounded five-factor value system tailored for LLMs. For systematic validation, we present three benchmarking tasks that integrate psychological principles with cutting-edge AI priorities. Our results reveal that the proposed value system meets standard psychological criteria, better captures LLM values, improves LLM safety prediction, and enhances LLM alignment, when compared to the canonical Schwartz's values.

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。

↑