Lecture | Leiden Translation Talks
Reducing or Reinforcing Gender Bias? A Study on the Application of ChatGPT in Translation from a Feminist Perspective
- Date
- Thursday 6 November 2025
- Time
- Series
- Leiden Translation Talks
- Location
-
Johan Huizinga
Doelensteeg 16
2311 VL Leiden - Room
- 0.09 (The Digital Lab)
Abstract
With AI increasingly involved in translation practice, its impact on gender bias in translation requires an investigation from the feminist perspective. This study examined 87 translation cases extracted from widespread social media discussions that expressed dissatisfaction with the gender bias in the translation. ChatGPT-4 was applied in four experiments: scoring the source texts for pre-translation preparation, translating 87 cases, and evaluating the translations by scoring the parallel texts, as well as the source and target texts respectively. The findings revealed that though ChatGPT’s identification of gender bias differed from opinions on social media, it demonstrated potential in reducing gender bias by detecting bias in the translation, remaining faithful to the source texts, employing inclusive expressions, or adding notes, especially when guided by proper prompts. Drawing on the feminist translation theory, this study attempted to offer feasible suggestions for reducing gender bias in translation in the era of AI.