Challenge: Recent studies show that textual entailment learning reduces social biases in pretrained sentence encoders.
Approach: They compare pretrained sentence encoders with textual entailment models that learn language logic for downstream language understanding tasks.
Outcome: The proposed models outperform models with lower bias without debiasing processes on stereotype, profession, and emotion bias tests.

Similar Papers

StereoSet: Measuring stereotypical bias in pretrained language models (2021.acl-long)

Copied to clipboard

Challenge: Existing literature on stereotypical biases in language models is limited . current evaluations focus on measuring bias without considering language modeling ability .
Approach: They propose to measure stereotypical biases in four domains: gender, profession, race, and religion . they compare stereotypical and language modeling ability of popular models like BERT, GPT-2, RoBERTa and XLnet .
Outcome: The proposed model shows strong stereotypical biases in gender, profession, race, and religion domains.
He is very intelligent, she is very beautiful? On Mitigating Social Biases in Language Modelling and Generation (2021.findings-acl)

Copied to clipboard

Challenge: Existing studies have focused on mitigating social biases in context-free representations, with recent shift to contextual ones.
Approach: They propose an approach to mitigate social biases in a large pre-trained contextual language model . they propose lexical co-occurrence-based bias penalization in the decoder units .
Outcome: The proposed approach reduces biases in fill-in-the-blank sentences and summarizes . it also reduces the biased representations in the frameworks, the authors show .
Towards Debiasing Sentence Representations (2020.acl-main)

Copied to clipboard

Challenge: Recent work has shown word-level embeddings reflect and propagate social biases present in training corpora.
Approach: They propose a method to debias word embeddings to reduce biases at sentence level . they hope their work will inspire future research on characterizing and removing biase .
Outcome: The proposed method reduces biases and preserves performance on downstream tasks such as sentiment analysis and natural language understanding.
Does Reasoning Introduce Bias? A Study of Social Bias Evaluation and Mitigation in LLM Reasoning (2025.findings-emnlp)

Copied to clipboard

Challenge: Recent advances in large language models have enabled automatic generation of chain-of-thought reasoning . however, when reasoning steps reflect social stereotypes, they can reinforce harmful associations and lead to misleading conclusions.
Approach: They propose a method that detects how model predictions change across incremental reasoning steps.
Outcome: The proposed method outperforms a stereotype-free baseline and improves accuracy.
Uncovering and Categorizing Social Biases in Text-to-SQL (2023.acl-long)

Copied to clipboard

Challenge: Existing Text-to-SQL models are trained on clean, neutral datasets, such as Spider and WikiSQl, but these models contain social bias at different rates.
Approach: They propose to use data to map natural language utterances to SQL queries.
Outcome: The proposed model can contain social bias at different rates in the downstream Text-to-SQL task.
From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Models (2023.acl-long)

Copied to clipboard

Challenge: Hundreds of studies have highlighted ethical issues in NLP models .
Approach: They propose to measure media biases in LMs trained on diverse data sources . they focus on hate speech and misinformation detection .
Outcome: The proposed methods quantify the fairness of downstream NLP models trained on politically biased LMs.
A Study of Implicit Bias in Pretrained Language Models against People with Disabilities (2022.coling-1)

Copied to clipboard

Challenge: Pretrained language models exhibit sociodemographic biases, such as against gender and race, raising concerns of downstream biase in language technologies.
Approach: They propose to use word embedding-based and transformer-based PLMs to test for the presence of biases against people with disabilities (PWDs)
Outcome: The proposed models favor ableist language, despite their sociodemographic biases against race and gender.
When Debiasing Backfires: Counterintuitive Side Effects of Preprocessing-Based Stereotype Mitigation (2026.findings-acl)

Copied to clipboard

Challenge: Preprocessing-based methods for stereotype mitigation are widely used in NLP . preprocessing methods cause unintended shifts in attention flow, authors say .
Approach: They propose to use preprocessing-based methods to reduce stereotypes for targeted groups . they find that stereotyping or counter-stereotyping can increase for other demographics .
Outcome: The proposed methods often induce unintended shifts across demographics, the authors show . they show that such side effects are not accompanied by large changes in attention flow .
Debiasing Pretrained Text Encoders by Paying Attention to Paying Attention (2022.emnlp-main)

Copied to clipboard

Challenge: Recent research has exposed text encoders for replicating discriminatory social biases which may cause unintended and undesired model behaviors with respect to social groups.
Approach: They propose a method to reduce social stereotypes by redistributing attention scores of a text encoder so it forgets any preference to historically advantaged groups and attends to all social classes with the same intensity.
Outcome: The proposed method reduces stereotypes and inflicts no semantic damage on pre-trained encoders.
When Do Pre-Training Biases Propagate to Downstream Tasks? A Case Study in Text Summarization (2023.eacl-main)

Copied to clipboard

Challenge: Existing studies have shown that large language models contain linguistic and societal biases, but it is unclear how these biase amplify to downstream tasks.
Approach: They investigate how name-nationality bias propagates from pre-training to downstream tasks . they show that these biases manifest themselves as hallucinations in summarization .
Outcome: The proposed model can reduce the rate of hallucinations, but does not change the types of biases that do appear.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations