Ethical Reasoning and Moral Value Alignment of LLMs Depend on the Language we Prompt them in

2024/04/29 by Utkarsh Agarwal, Agarwal, Utkarsh, Kumar Tanmay +5 · 19 citations
Business, Management and Accounting · Decision Sciences · Social Sciences · #Artificial Intelligence (cs.AI) #Business Law and Ethics #Computation and Language (cs.CL) #Ethics in Business and Education #FOS: Computer and information sciences #Legal Education and Practice Innovations

paper · pdf · doi:10.48550/arxiv.2404.18460

openalex publication_date 2024/04/29 · openalex created_date 2024/05/11 · openalex updated_date 2026/07/28

Abstract

Ethical reasoning is a crucial skill for Large Language Models (LLMs). However, moral values are not universal, but rather influenced by language and culture. This paper explores how three prominent LLMs -- GPT-4, ChatGPT, and Llama2-70B-Chat -- perform ethical reasoning in different languages and if their moral judgement depend on the language in which they are prompted. We extend the study of ethical reasoning of LLMs by Rao et al. (2023) to a multilingual setup following their framework of probing LLMs with ethical dilemmas and policies from three branches of normative ethics: deontology, virtue, and consequentialism. We experiment with six languages: English, Spanish, Russian, Chinese, Hindi, and Swahili. We find that GPT-4 is the most consistent and unbiased ethical reasoner across languages, while ChatGPT and Llama2-70B-Chat show significant moral value bias when we move to languages other than English. Interestingly, the nature of this bias significantly vary across languages for all LLMs, including GPT-4.

Cited by

• Made-in China, Thinking in America:U.S. Values Persist in Chinese LLMs
• Cross-cultural value alignment frameworks for responsible AI governance: Evidence from China-West comparative analysis
• EvalMORAAL: Interpretable Chain-of-Thought and LLM-as-Judge Evaluation for Moral Alignment in Large Language Models
• How do LLMs reflect human moral foundations? a study using the moral foundations framework
• Pluralistic Alignment for Healthcare: A Role-Driven Framework
• What if I ask in alia lingua? Measuring Functional Similarity Across Languages
• Dynamite: Real-Time Debriefing Slide Authoring through AI-Enhanced Multimodal Interaction
• Black Box Deployed -- Functional Criteria for Artificial Moral Agents in the LLM Era
• EtiCor++: Towards Understanding Etiquettical Bias in LLMs
• "Amazing, They All Lean Left" -- Analyzing the Political Temperaments of Current LLMs
• MFTCXplain: A Multilingual Benchmark Dataset for Evaluating the Moral Reasoning of LLMs through Multi-hop Hate Speech Explanation
• A Game-Theoretic Negotiation Framework for Cross-Cultural Consensus in LLMs
• Exploring Cultural Variations in Moral Judgments with Large Language Models
• Do Large Language Models Know Folktales? A Case Study of Yokai in Japanese Folktales
• MAKIEval: A Multilingual Automatic WiKidata-based Framework for Cultural Awareness Evaluation for LLMs
• Charting the Landscape of African NLP: Mapping Progress and Shaping the Road Ahead
• Analyzing Political Bias in LLMs via Target-Oriented Sentiment Classification
• OpenEthics: A Comprehensive Ethical Evaluation of Open-Source Generative Large Language Models
• Uncovering Cultural Representation Disparities in Vision-Language Models

Related

• Do Moral Judgment and Reasoning Capability of LLMs Change with Language? A Study using the Multilingual Defining Issues Test
• Moral Reasoning Across Languages: The Critical Role of Low-Resource Languages in LLMs
• Decoding Multilingual Moral Preferences: Unveiling LLM's Biases Through the Moral Machine Experiment
• One Model, Many Morals: Uncovering Cross-Linguistic Misalignments in Computational Moral Reasoning
• Whose Morality Do They Speak? Unraveling Cultural Bias in Multilingual Language Models