Challenge: Large Language Models (LLMs) can learn useful knowledge and harmful stereotypes, making bias evaluation essential.
Approach: They propose a multilingual social bias benchmark that incorporates human-generated reasoning as part of the thinking process.
Outcome: The proposed method demonstrates superior performance over LLM-generated methods . human-generated thinking yields higher-quality evaluations than template-based approaches .

Similar Papers

Mind the Gap: Multilingual Divide in LLM Bias Detection and Reasoning (2026.acl-srw)

Copied to clipboard

Challenge: Large Language Models (LLMs) are increasingly deployed in multilingual settings . but most bias evaluation remains English-centric and ignores how bias manifests within reasoning .
Approach: They evaluate large language models with supervised fine-tuning and preference optimization . they find that bias varies substantially across languages, with consistent degradation in non-English settings .
Outcome: The proposed model improves in English, Dutch, Spanish, and Turkish using the MBBQ benchmark.
Social Bias in Multilingual Language Models: A Survey (2025.emnlp-main)

Copied to clipboard

Challenge: Pretrained multilingual models exhibit the same social bias as models processing English texts.
Approach: They examine the literature on bias evaluation and mitigation approaches in multilingual and non-English contexts and identify gaps in the field.
Outcome: The proposed models perform well on multilingual language understanding benchmarks and are consistent with the current literature.
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.
Social Bias Benchmark for Generation: A Comparison of Generation and QA-Based Evaluations (2025.findings-acl)

Copied to clipboard

Challenge: Existing methods for assessing social bias in large language models (LLMs) do not capture nuanced and context-dependent nature of natural language generation.
Approach: They propose a Bias Benchmark for Generation (BBG) that evaluates social bias in long-form generation by having LLMs generate continuations of story prompts.
Outcome: The proposed benchmark is based on the English BBQ and Korean BBQ datasets and compares it with multiplechoice BBQ evaluation.
Multi-Persona Thinking for Bias Mitigation in Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Large Language Models exhibit social biases, which can lead to harmful stereotypes and unfair outcomes.
Approach: They propose a simple inference-time framework that encourages reasoning from multiple perspectives.
Outcome: The proposed framework reduces bias by encouraging reasoning from multiple perspectives.
7 Points to Tsinghua but 10 Points to ? Assessing Large Language Models in Agentic Multilingual National Bias (2025.findings-acl)

Copied to clipboard

Challenge: Large Language Models have garnered significant attention for their capabilities in multilingual natural language processing, but studies on risks associated with cross biases are limited to immediate context preferences.
Approach: They investigate multilingual bias in state-of-the-art Large Language Models by analyzing their responses to decision-making tasks across multiple languages.
Outcome: The proposed model can provide personalized advice across university applications, travel, and relocation scenarios.
TWBias: A Benchmark for Assessing Social Bias in Traditional Chinese Large Language Models through a Taiwan Cultural Lens (2024.findings-emnlp)

Copied to clipboard

Challenge: Large language models have shown remarkable capabilities in natural language processing, but concerns about social bias amplification remain.
Approach: They propose a social bias evaluation benchmark for Traditional Chinese LLMs that integrates chat templates and diverse prompts for comprehensive bias assessment.
Outcome: The proposed model incorporates chat templates and diverse prompts for comprehensive bias assessment focusing on Taiwan's cultural context and prioritizing gender and ethnicity bias evaluation.
Your Stereotypical Mileage May Vary: Practical Challenges of Evaluating Biases in Multiple Languages and Cultural Contexts (2024.lrec-main)

Copied to clipboard

Challenge: Recent studies have identified a gap in the availability of tools and resources to study bias in languages other than English and social contexts outside the north of America.
Approach: They use stereotypes to build a corpus of sentence pairs that cover biases in seven cultural contexts.
Outcome: The proposed resource covers a wide range of languages and cultural settings . it favors sentences that express stereotypes in most bias categories .
Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation (2025.acl-long)

Copied to clipboard

Challenge: Reliable multilingual evaluation is difficult and culturally appropriate evaluation is even harder to achieve.
Approach: They propose a multilingual evaluation framework that aims to mitigate these biases by improving translations and annotation practices.
Outcome: The proposed framework improves translation quality and cultural coverage and is culturally sensitive and culturally agnostic.
Rethinking Pragmatics in Large Language Models: Towards Open-Ended Evaluation and Preference Tuning (2024.emnlp-main)

Copied to clipboard

Challenge: Existing methods to assess social-pragmatic inference in large language models are inadequacy, and preferential tuning is the best approach.
Approach: They propose to use free-form models' responses as a measure to assess social-pragmatic reasoning and advocate for preference optimization over supervised finetuning (SFT).
Outcome: The proposed model outperforms supervised finetuning (SFT) and offers a near-free launch in pragmatic abilities without compromising general capabilities.

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