# Redirected, Not Removed: Task-Dependent Stereotyping Reveals the Limits of LLM Alignments arXiv: https://arxiv.org/abs/2604.02669 ยท HTML: https://mlsanity.com/paper/2604.02669/ ## Key facts - Cited by 1 papers in the map (data through 14 Oct 2026). - Placed in Human values and AI feedback alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation. - Builds on 32 of its 42 references that are in the map. - First posted on arXiv 3 Apr 2026. ## Placed in - LLM alignment / RLHF / preference optimisation / [Human values and AI feedback alignment](https://mlsanity.com/alignment/human-values-and-ai-feedback-alignment/) ## Builds on (most cited first, 30 of 32) - [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023) - [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022) - [RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models](https://mlsanity.com/paper/2009.11462/) (2020) - [StereoSet: Measuring stereotypical bias in pretrained language models](https://mlsanity.com/paper/2004.09456/) (2020) - [Semantics derived automatically from language corpora contain human-like biases](https://mlsanity.com/paper/1608.07187/) (2016) - [ChatGPT Outperforms Crowd-Workers for Text-Annotation Tasks](https://mlsanity.com/paper/2303.15056/) (2023) - [BBQ: A Hand-Built Bias Benchmark for Question Answering](https://mlsanity.com/paper/2110.08193/) (2021) - [Gender Bias in Coreference Resolution: Evaluation and Debiasing Methods](https://mlsanity.com/paper/1804.06876/) (2018) - [Bias and Fairness in Large Language Models: A Survey](https://mlsanity.com/paper/2309.00770/) (2023) - [XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models](https://mlsanity.com/paper/2308.01263/) (2023) - [Gender Bias in Coreference Resolution](https://mlsanity.com/paper/1804.09301/) (2018) - [BOLD: Dataset and Metrics for Measuring Biases in Open-Ended Language Generation](https://mlsanity.com/paper/2101.11718/) (2021) - [Detecting Emergent Intersectional Biases: Contextualized Word Embeddings Contain a Distribution of Human-like Biases](https://mlsanity.com/paper/2006.03955/) (2020) - [Large Language Models are Geographically Biased](https://mlsanity.com/paper/2402.02680/) (2024) - [Fairness And Bias in Artificial Intelligence: A Brief Survey of Sources, Impacts, And Mitigation Strategies](https://mlsanity.com/paper/2304.07683/) (2023) - [BiasAsker: Measuring the Bias in Conversational AI System](https://mlsanity.com/paper/2305.12434/) (2023) - [Measuring Implicit Bias in Explicitly Unbiased Large Language Models](https://mlsanity.com/paper/2402.04105/) (2024) - [Probing Explicit and Implicit Gender Bias through LLM Conditional Text Generation](https://mlsanity.com/paper/2311.00306/) (2023) - [Large Language Models Portray Socially Subordinate Groups as More Homogeneous, Consistent with a Bias Observed in Humans](https://mlsanity.com/paper/2401.08495/) (2024) - [Explicit vs. Implicit: Investigating Social Bias in Large Language Models through Self-Reflection](https://mlsanity.com/paper/2501.02295/) (2025) - [Investigating Implicit Bias in Large Language Models: A Large-Scale Study of Over 50 LLMs](https://mlsanity.com/paper/2410.12864/) (2024) - [Large Language Models Still Exhibit Bias in Long Text](https://mlsanity.com/paper/2410.17519/) (2024) - [''Fifty Shades of Bias'': Normative Ratings of Gender Bias in GPT Generated English Text](https://mlsanity.com/paper/2310.17428/) (2023) - [How Deep Is Representational Bias in LLMs? The Cases of Caste and Religion](https://mlsanity.com/paper/2508.03712/) (2025) - [Social Bias in Large Language Models For Bangla: An Empirical Study on Gender and Religious Bias](https://mlsanity.com/paper/2407.03536/) (2024) - [Divine LLaMAs: Bias, Stereotypes, Stigmatization, and Emotion Representation of Religion in Large Language Models](https://mlsanity.com/paper/2407.06908/) (2024) - [DECASTE: Unveiling Caste Stereotypes in Large Language Models through Multi-Dimensional Bias Analysis](https://mlsanity.com/paper/2505.14971/) (2025) - [FairI Tales: Evaluation of Fairness in Indian Contexts with a Focus on Bias and Stereotypes](https://mlsanity.com/paper/2506.23111/) (2025) - [Bias in Large Language Models Across Clinical Applications: A Systematic Review](https://mlsanity.com/paper/2504.02917/) (2025) - [HInter: Exposing Hidden Intersectional Bias in Large Language Models](https://mlsanity.com/paper/2503.11962/) (2025) ## Built on by (most cited first, 1 of 1) - [The Attentional White Bear Effect in Transformer Language Models](https://mlsanity.com/paper/2605.28639/) (2026) ## Abstract (arXiv) How biased is a language model? The answer depends on how you ask. A model that refuses to choose between castes for a leadership role will, in a fill-in-the-blank task, reliably associate upper castes with purity and lower castes with lack of hygiene. Single-task benchmarks miss this because they capture only one slice of a model's bias profile. We introduce a hierarchical taxonomy covering 9 bias types, including under-studied axes like caste, linguistic, and geographic bias, operationalized through 7 evaluation tasks that span explicit decision-making to implicit association. Auditing 7 commercial and open-weight LLMs with textasciitilde45K prompts, we find three systematic patterns. First, bias is task-dependent: models counter stereotypes on explicit probes but reproduce them on implicit ones, with Stereotype Score divergences up to 0.43 between task types for the same model and identity groups. Second, safety alignment is asymmetric: models refuse to assign negative traits to marginalized groups, but freely associate positive traits with privileged ones. Third, under-studied bias axes show the strongest stereotyping across all models, suggesting alignment effort tracks benchmark coverage rather than harm severity. These results demonstrate that single-benchmark audits systematically mischaracterize LLM bias and that current alignment practices mask representational harm rather than mitigating it.