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

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ye, Haoran, Zhang, Tianze, Xie, Yuhang, Zhang, Liyuan, Ren, Yuanyi, Zhang, Xin, Song, Guojie
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909829392498688
author Ye, Haoran
Zhang, Tianze
Xie, Yuhang
Zhang, Liyuan
Ren, Yuanyi
Zhang, Xin
Song, Guojie
author_facet Ye, Haoran
Zhang, Tianze
Xie, Yuhang
Zhang, Liyuan
Ren, Yuanyi
Zhang, Xin
Song, Guojie
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2502_02444
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Generative Psycho-Lexical Approach for Constructing Value Systems in Large Language Models
Ye, Haoran
Zhang, Tianze
Xie, Yuhang
Zhang, Liyuan
Ren, Yuanyi
Zhang, Xin
Song, Guojie
Computation and Language
Artificial Intelligence
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.
title Generative Psycho-Lexical Approach for Constructing Value Systems in Large Language Models
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2502.02444