BiasGym is a cost‑effective, generalizable framework that injects specific biases into large language models via token‑based fine‑tuning while keeping the model frozen. It then uses two debiasing methods—Scope and Steer—to identify and suppress or redirect the components responsible for biased behavior. The framework enables consistent bias elicitation, precise localization of bias associations, and targeted debiasing without harming downstream performance, and it has been shown to reduce real‑world stereotypes such as labeling Italians as reckless drivers.
By Sekh Mainul Islam, Nadav Borenstein, Siddhesh Milind Pawar, Haeun Yu, Arnav Arora, Isabelle Augenstein
arXiv:2311.13892v4 Announce Type: replace-cross
Abstract: The social biases and unwelcome stereotypes revealed by pretrained language models are becoming obstacles to their application. Compared to n...
By Bingkang Shi, Xiaodan Zhang, Dehan Kong, Yulei Wu, Zongzhen Liu, Honglei Lyu, Longtao Huang
arXiv:2608. 14161v1 Announce Type: new Abstract: LLMs exhibit social biases that can produce inaccurate and discriminatory inferences, posing risks in high-stakes applications.
By Varsha Ramineni, Hossein A. Rahmani, Jerome Ramos, Karin Sevegnani, Emine Yilmaz
The paper presents GRASP, a method for unsupervised detection and removal of spurious correlations that arise during fine‑tuning of pretrained language models. By identifying latent factors from LoRA fine‑tune weights and projecting gradients to prevent new reliance on them, GRASP preserves genuine task signals while mitigating bias. Experiments on three tasks—misaligned code generation, medical advice, and political bias—show that GRASP reduces misalignment and drift more effectively than existing baselines while maintaining task performance.
By Ciar\'an M. Gilligan-Lee, Joseph Egan, Yuchen Zhu, Michael O'Riordan
The paper examines how large language models (LLMs) can be biased by irrelevant social contexts when evaluating teachers, using a large U.S. classroom transcript dataset. It shows that spurious contexts can shift model ratings by up to 1.48 points on a 7‑point scale and that standard mitigation methods like SFT and DPO are insufficient. The authors introduce Debiasing‑DPO, a method that combines contrastive reasoning‑augmented DPO with SFT, which reduces bias by 84% and improves predictive accuracy by 52% on Llama and Qwen Instruct models.
By Hyunji Nam, Dorottya Demszky
The paper introduces JudgeBiasBench, a benchmark that systematically quantifies judgment biases in large language model (LLM)-based judges across four dimensions and 12 bias types. It evaluates both generative and discriminative judges, revealing significant bias patterns that undermine reliability. The authors propose bias-aware training—reinforcement learning for generative judges and contrastive learning for discriminative judges—to reduce these biases while maintaining evaluation performance.
By Hongli Zhou, Hui Huang, Rui Zhang, Kehai Chen, Bing Xu, Conghui Zhu, Tiejun Zhao, Muyun Yang
arXiv:2604. 01925v2 Announce Type: replace-cross Abstract: Large Language Models increasingly suppress biased outputs when demographic identity is stated explicitly, yet may still exhibit implicit biases when identity is conveyed indirectly.
By Bhaskara Hanuma Vedula, Darshan Anghan, Ishita Goyal, Ponnurangam Kumaraguru, Abhijnan Chakraborty
The paper introduces GPTBIAS, a framework that uses powerful large language models like GPT‑4 to evaluate bias in other LLMs. It employs specially crafted prompts called Bias Attack Instructions to probe for bias and outputs a bias score along with detailed information such as bias types, affected demographics, keywords, reasons, and improvement suggestions. Extensive experiments demonstrate the framework’s effectiveness and usability.
By Jiaxu Zhao, Meng Fang, Shirui Pan, Wenpeng Yin, Mykola Pechenizkiy
arXiv:2607. 28319v1 Announce Type: cross Abstract: This work presents Fairness Pruning, a lightweight structural intervention method designed for the management and future mitigation of demographic bias in large language models (LLMs).
By Pere Martra, Eugenio Mart\'inez C\'amara, Alfonso Ure\~na L\'opez
The paper investigates whether language models still encode occupational biases even when they appear unbiased in behavioral tests. Using a causal framework, the authors separate bias into internal representations of user competence and observable outputs, deriving steering vectors that show these representations influence model behavior in question‑answering and hiring tasks. Across several open‑weight models, demographic factors such as gender, race, and socioeconomic status affect the models’ internal competence representations, revealing hidden bias that behavioral metrics alone may miss.
By Keren Fuentes, Aaron Mueller
arXiv:2607. 05679v1 Announce Type: cross Abstract: Language models (LMs) exhibit problematic biases, such as stereotypes.
By Damian Hodel, Jevin West, Aylin Caliskan
arXiv:2601. 21864v2 Announce Type: replace Abstract: Large language models (LLMs) exhibit social biases that reinforce harmful stereotypes, limiting their safe deployment.
By Jinhao Pan, Chahat Raj, Anjishnu Mukherjee, Sina Mansouri, Bowen Wei, Shloka Yada, Ziwei Zhu