The power of Prompts: Evaluating and Mitigating Gender Bias in MT with LLMs

TL;DR

A prompt structure is identified that significantly reduces gender bias by up to 12% on the WinoMT evaluation dataset compared to more straightforward prompts, which significantly reduce the gender bias accuracy gap between LLMs and traditional NMT systems.

Abstract

This paper studies gender bias in machine translation through the lens of Large Language Models (LLMs). Four widely-used test sets are employed to benchmark various base LLMs, comparing their translation quality and gender bias against state-of-the-art Neural Machine Translation (NMT) models for English to Catalan (En $\rightarrow$ Ca) and English to Spanish (En $\rightarrow$ Es) translation directions. Our findings reveal pervasive gender bias across all models, with base LLMs exhibiting a higher degree of bias compared to NMT models. To combat this bias, we explore prompting engineering techniques applied to an instruction-tuned LLM. We identify a prompt structure that significantly reduces gender bias by up to 12% on the WinoMT evaluation dataset compared to more straightforward prompts. These results significantly reduce the gender bias accuracy gap between LLMs and traditional NMT systems.

Figures (4)

• Figure 1: Example of Gender Bias in MT
• Figure 2: Examples of Gender Coreference Resolution (a) and Gender Terms Detection (b) in En $\rightarrow$ Ca
• Figure 3: Male and female predicted terms across models for En $\rightarrow$ Ca in absence of gender cues
• Figure 4: Male and female predicted terms across models for En $\rightarrow$ Es in absence of gender cues