# Discrimination by LLMs: Cross-lingual Bias Assessment and Mitigation in Decision-Making and Summarisation

arXiv: https://arxiv.org/abs/2509.09735 · HTML: https://mlsanity.com/paper/2509.09735/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 35 references that are in the map.
- First posted on arXiv 10 Sep 2025.

## Builds on (most cited first, 22 of 22)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://mlsanity.com/paper/1910.13461/) (2019)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formatting](https://mlsanity.com/paper/2310.11324/) (2023)
- [A Comprehensive Overview of Large Language Models](https://mlsanity.com/paper/2307.06435/) (2023)
- [Gender Bias in Contextualized Word Embeddings](https://mlsanity.com/paper/1904.03310/) (2019)
- [Evaluating and Mitigating Discrimination in Language Model Decisions](https://mlsanity.com/paper/2312.03689/) (2023)
- [Reducing Gender Bias in Word-Level Language Models with a Gender-Equalizing Loss Function](https://mlsanity.com/paper/1905.12801/) (2019)
- [Determinants of LLM-assisted Decision-Making](https://mlsanity.com/paper/2402.17385/) (2024)
- [Comparing Biases and the Impact of Multilingual Training across Multiple Languages](https://mlsanity.com/paper/2305.11242/) (2023)
- [Are We Consistently Biased? Multidimensional Analysis of Biases in Distributional Word Vectors](https://mlsanity.com/paper/1904.11783/) (2019)
- [The Silicon Ceiling: Auditing GPT's Race and Gender Biases in Hiring](https://mlsanity.com/paper/2405.04412/) (2024)
- [Looking for a Handsome Carpenter! Debiasing GPT-3 Job Advertisements](https://mlsanity.com/paper/2205.11374/) (2022)
- [Bias in News Summarization: Measures, Pitfalls and Corpora](https://mlsanity.com/paper/2309.08047/) (2023)
- [This is not a Dataset: A Large Negation Benchmark to Challenge Large Language Models](https://mlsanity.com/paper/2310.15941/) (2023)
- [Gene Set Summarization using Large Language Models](https://mlsanity.com/paper/2305.13338/) (2023)
- [Bias in Opinion Summarisation from Pre-training to Adaptation: A Case Study in Political Bias](https://mlsanity.com/paper/2402.00322/) (2024)
- [The power of Prompts: Evaluating and Mitigating Gender Bias in MT with LLMs](https://mlsanity.com/paper/2407.18786/) (2024)
- [Leveraging Large Language Model to Generate a Novel Metaheuristic Algorithm with CRISPE Framework](https://mlsanity.com/paper/2403.16417/) (2024)
- [Mitigating Bias in Conversations: A Hate Speech Classifier and Debiaser with Prompts](https://mlsanity.com/paper/2307.10213/) (2023)
- [The Extractive-Abstractive Axis: Measuring Content "Borrowing" in Generative Language Models](https://mlsanity.com/paper/2307.11779/) (2023)

## Abstract (arXiv)

The rapid integration of Large Language Models (LLMs) into various domains raises concerns about societal inequalities and information bias. This study examines biases in LLMs related to background, gender, and age, with a focus on their impact on decision-making and summarization tasks. Additionally, the research examines the cross-lingual propagation of these biases and evaluates the effectiveness of prompt-instructed mitigation strategies. Using an adapted version of the dataset by Tamkin et al. (2023) translated into Dutch, we created 151,200 unique prompts for the decision task and 176,400 for the summarisation task. Various demographic variables, instructions, salience levels, and languages were tested on GPT-3.5 and GPT-4o. Our analysis revealed that both models were significantly biased during decision-making, favouring female gender, younger ages, and certain backgrounds such as the African-American background. In contrast, the summarisation task showed minimal evidence of bias, though significant age-related differences emerged for GPT-3.5 in English. Cross-lingual analysis showed that bias patterns were broadly similar between English and Dutch, though notable differences were observed across specific demographic categories. The newly proposed mitigation instructions, while unable to eliminate biases completely, demonstrated potential in reducing them. The most effective instruction achieved a 27% mean reduction in the gap between the most and least favorable demographics. Notably, contrary to GPT-3.5, GPT-4o displayed reduced biases for all prompts in English, indicating the specific potential for prompt-based mitigation within newer models. This research underscores the importance of cautious adoption of LLMs and context-specific bias testing, highlighting the need for continued development of effective mitigation strategies to ensure responsible deployment of AI.