Paper Type
ERF
Abstract
This paper explores the ability of large language models (LLMs) to understand and respect users' philosophical positions. It looks back at the problems of bias and misleading information in LLM applications and puts forward two research questions: How well do LLMs understand and respect different philosophical positions when giving advice? Do LLMs adjust their responses based on users' philosophical positions or do they start from a specific philosophical perspective by default? The study collected data on major philosophical ideas and their core values, created a corpus of prompts with different scenarios, and analyzed the LLMs' responses. Through systematic evaluation, the study aims to provide guidance for LLM developers and enrich the theoretical system of the intersection between philosophy and artificial intelligence.
Paper Number
1629
Recommended Citation
LI, SHAN; Qi, Jiayin; and chen, kaifan, "To What Extent Do LLMs Respect the Philosophical Stance of Users?" (2025). AMCIS 2025 Proceedings. 9.
https://aisel.aisnet.org/amcis2025/intelfuture/intelfuture/9
To What Extent Do LLMs Respect the Philosophical Stance of Users?
This paper explores the ability of large language models (LLMs) to understand and respect users' philosophical positions. It looks back at the problems of bias and misleading information in LLM applications and puts forward two research questions: How well do LLMs understand and respect different philosophical positions when giving advice? Do LLMs adjust their responses based on users' philosophical positions or do they start from a specific philosophical perspective by default? The study collected data on major philosophical ideas and their core values, created a corpus of prompts with different scenarios, and analyzed the LLMs' responses. Through systematic evaluation, the study aims to provide guidance for LLM developers and enrich the theoretical system of the intersection between philosophy and artificial intelligence.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
IntelFuture