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

TL;DR

This study investigates how the ethical reasoning of LLMs varies across different languages, revealing that GPT-4 maintains consistency while others exhibit significant language-dependent moral biases.

Contribution

It extends prior ethical reasoning research to a multilingual context, analyzing how language influences moral judgments in LLMs across multiple ethical frameworks.

Findings

GPT-4 shows consistent ethical reasoning across languages.

ChatGPT and Llama2-70B-Chat exhibit language-dependent moral biases.

Biases vary significantly across languages for all tested LLMs.

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.