# Investigating Social Bias Changes in Quantized Language Models

arXiv: https://arxiv.org/abs/2602.06181 · HTML: https://mlsanity.com/paper/2602.06181/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 30 of its 43 references that are in the map.
- First posted on arXiv 5 Feb 2026.

## Builds on (most cited first, 30 of 30)

- [GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers](https://mlsanity.com/paper/2210.17323/) (2022)
- [Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference](https://mlsanity.com/paper/1712.05877/) (2017)
- [SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models](https://mlsanity.com/paper/2211.10438/) (2022)
- [Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings](https://mlsanity.com/paper/1607.06520/) (2016)
- [AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration](https://mlsanity.com/paper/2306.00978/) (2023)
- [SimPO: Simple Preference Optimization with a Reference-Free Reward](https://mlsanity.com/paper/2405.14734/) (2024)
- [StereoSet: Measuring stereotypical bias in pretrained language models](https://mlsanity.com/paper/2004.09456/) (2020)
- [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)
- [BBQ: A Hand-Built Bias Benchmark for Question Answering](https://mlsanity.com/paper/2110.08193/) (2021)
- [Bias and Fairness in Large Language Models: A Survey](https://mlsanity.com/paper/2309.00770/) (2023)
- [Large Language Models Are Not Robust Multiple Choice Selectors](https://mlsanity.com/paper/2309.03882/) (2023)
- [Large Language Models Sensitivity to The Order of Options in Multiple-Choice Questions](https://mlsanity.com/paper/2308.11483/) (2023)
- [Intrinsic Bias Metrics Do Not Correlate with Application Bias](https://mlsanity.com/paper/2012.15859/) (2020)
- [The Dawn After the Dark: An Empirical Study on Factuality Hallucination in Large Language Models](https://mlsanity.com/paper/2401.03205/) (2024)
- [Catastrophic Failure of LLM Unlearning via Quantization](https://mlsanity.com/paper/2410.16454/) (2024)
- [A Peek into Token Bias: Large Language Models Are Not Yet Genuine Reasoners](https://mlsanity.com/paper/2406.11050/) (2024)
- [Decoding Compressed Trust: Scrutinizing the Trustworthiness of Efficient LLMs Under Compression](https://mlsanity.com/paper/2403.15447/) (2024)
- [Measuring Implicit Bias in Explicitly Unbiased Large Language Models](https://mlsanity.com/paper/2402.04105/) (2024)
- [Debiasing isn't enough! -- On the Effectiveness of Debiasing MLMs and their Social Biases in Downstream Tasks](https://mlsanity.com/paper/2210.02938/) (2022)
- [FairMT-Bench: Benchmarking Fairness for Multi-turn Dialogue in Conversational LLMs](https://mlsanity.com/paper/2410.19317/) (2024)
- [Beyond Perplexity: Multi-dimensional Safety Evaluation of LLM Compression](https://mlsanity.com/paper/2407.04965/) (2024)
- [ChatBug: A Common Vulnerability of Aligned LLMs Induced by Chat Templates](https://mlsanity.com/paper/2406.12935/) (2024)
- [Choose Your Lenses: Flaws in Gender Bias Evaluation](https://mlsanity.com/paper/2210.11471/) (2022)
- [Unlocking Tokens as Data Points for Generalization Bounds on Larger Language Models](https://mlsanity.com/paper/2407.18158/) (2024)
- [Understanding the Effect of Model Compression on Social Bias in Large Language Models](https://mlsanity.com/paper/2312.05662/) (2023)
- [CEB: Compositional Evaluation Benchmark for Fairness in Large Language Models](https://mlsanity.com/paper/2407.02408/) (2024)
- [SocialStigmaQA: A Benchmark to Uncover Stigma Amplification in Generative Language Models](https://mlsanity.com/paper/2312.07492/) (2023)
- [Do Compressed LLMs Forget Knowledge? An Experimental Study with Practical Implications](https://mlsanity.com/paper/2310.00867/) (2023)
- [When Quantization Affects Confidence of Large Language Models?](https://mlsanity.com/paper/2405.00632/) (2024)
- [How Quantization Shapes Bias in Large Language Models](https://mlsanity.com/paper/2508.18088/) (2025)

## Built on by (most cited first, 5 of 5)

- [Quantized Reasoning Models Think They Need to Think Longer, but They Do Not](https://mlsanity.com/paper/2606.00206/) (2026)
- [Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels](https://mlsanity.com/paper/2605.15208/) (2026)
- [Weight Pruning Amplifies Bias: A Multi-Method Study of Compressed LLMs for Edge AI](https://mlsanity.com/paper/2605.08137/) (2026)
- [QuantiBias: Benchmarking Quantization-Induced Bias in LLMs](https://mlsanity.com/paper/2607.21063/) (2026)
- [Off-Target Effects of Response-Style Alignment in a Korean 27B Language Model](https://mlsanity.com/paper/2609.11291/) (2026)

## Abstract (arXiv)

Post-training quantization reduces the memory needed to run large language models but alters their social biases in ways that aggregate metrics fail to capture. We present the first large-scale study of 50 quantized models evaluated on PostTrainingBiasBench, a unified benchmark of 13 closed- and open-ended bias datasets. We identify a phenomenon we term quantization-induced bias flipping, in which quantization causes models to change responses from biased to unbiased and vice versa, up to 21% of the time, despite no change in aggregate bias scores. These flips are strongly associated with model uncertainty, where the responses with high uncertainty are 3-11x more likely to change than the confident ones. Quantization strength amplifies this effect, with 4-bit quantized models exhibiting 4-6x more behavioral changes than 8-bit quantized models. Critically, these changes create asymmetric impacts across demographic groups, where bias can worsen by up to 18.6% for some groups while improving by 14.1% for others, yielding misleadingly neutral aggregate outcomes. Larger models show no consistent robustness advantage, and group-specific shifts vary unpredictably across model families. Our findings demonstrate that compression fundamentally alters bias patterns, requiring crucial post-quantization evaluation and interventions to ensure reliability in practice.