Cognitive bias in high-stakes decision-making with llms

J Echterhoff, Y Liu, A Alessa, J McAuley… - arXiv preprint arXiv …, 2024 - arxiv.org
Large language models (LLMs) offer significant potential as tools to support an expanding
range of decision-making tasks. However, given their training on human (created) data …

Cognitive bias in decision-making with LLMs

J Echterhoff, Y Liu, A Alessa, J McAuley… - Findings of the …, 2024 - aclanthology.org
Large language models (LLMs) offer significant potential as tools to support an expanding
range of decision-making tasks. Given their training on human (created) data, LLMs have …

Metrics for What, Metrics for Whom: Assessing Actionability of Bias Evaluation Metrics in NLP

P Delobelle, G Attanasio, D Nozza… - Proceedings of the …, 2024 - aclanthology.org
This paper introduces the concept of actionability in the context of bias measures in natural
language processing (NLP). We define actionability as the degree to which a measure's …

The devil is in the neurons: Interpreting and mitigating social biases in language models

Y Liu, Y Liu, X Chen, PY Chen, D Zan… - The Twelfth …, 2024 - openreview.net
Pre-trained Language models (PLMs) have been acknowledged to contain harmful
information, such as social biases, which may cause negative social impacts or even bring …

A Survey on Multilingual Large Language Models: Corpora, Alignment, and Bias

Y Xu, L Hu, J Zhao, Z Qiu, Y Ye, H Gu - arXiv preprint arXiv:2404.00929, 2024 - arxiv.org
Based on the foundation of Large Language Models (LLMs), Multilingual Large Language
Models (MLLMs) have been developed to address the challenges of multilingual natural …

Cross-lingual Transfer Can Worsen Bias in Sentiment Analysis

S Goldfarb-Tarrant, B Ross, A Lopez - arXiv preprint arXiv:2305.12709, 2023 - arxiv.org
Sentiment analysis (SA) systems are widely deployed in many of the world's languages, and
there is well-documented evidence of demographic bias in these systems. In languages …

Evaluation of Attribution Bias in Retrieval-Augmented Large Language Models

A Abolghasemi, L Azzopardi, SH Hashemi… - arXiv preprint arXiv …, 2024 - arxiv.org
Attributing answers to source documents is an approach used to enhance the verifiability of
a model's output in retrieval augmented generation (RAG). Prior work has mainly focused on …

BiasWipe: Mitigating Unintended Bias in Text Classifiers through Model Interpretability

M Mamta, R Chigrupaatii, A Ekbal - Proceedings of the 2024 …, 2024 - aclanthology.org
Toxic content detection plays a vital role in addressing the misuse of social media platforms
to harm people or groups due to their race, gender or ethnicity. However, due to the nature …

Mitigating social bias in sentiment classification via ethnicity-aware algorithmic design

R Corizzo, FS Hafner - Social Network Analysis and Mining, 2024 - Springer
Sentiment analysis tools are frequently employed to analyze large amounts of natural
language data gathered from social networks and generate valuable insights on public …

MAFIA: Multi-Adapter Fused Inclusive LanguAge Models

P Jain, A Sathe, V Gumma, K Ahuja… - arXiv preprint arXiv …, 2024 - arxiv.org
Pretrained Language Models (PLMs) are widely used in NLP for various tasks. Recent
studies have identified various biases that such models exhibit and have proposed methods …