When Names Change Verdicts: Intervention Consistency Reveals Systematic Bias in LLM Decision-Making

• URL: http://arxiv.org/abs/2603.18530v1
• Date: Thu, 19 Mar 2026 06:21:08 GMT
• Title: When Names Change Verdicts: Intervention Consistency Reveals Systematic Bias in LLM Decision-Making
• Abstract summary: ICE-Guard is a framework applying intervention consistency testing to detect three types of spurious feature reliance.<n>We evaluate 11 large language models (LLMs) across 10 high-stakes domains.<n>We demonstrate an ICE-guided detect-diagnose-mitigate-verify loop achieving cumulative 78% bias reduction.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) are increasingly used for high-stakes decisions, yet their susceptibility to spurious features remains poorly characterized. We introduce ICE-Guard, a framework applying intervention consistency testing to detect three types of spurious feature reliance: demographic (name/race swaps), authority (credential/prestige swaps), and framing (positive/negative restatements). Across 3,000 vignettes spanning 10 high-stakes domains, we evaluate 11 LLMs from 8 families and find that (1) authority bias (mean 5.8%) and framing bias (5.0%) substantially exceed demographic bias (2.2%), challenging the field's narrow focus on demographics; (2) bias concentrates in specific domains -- finance shows 22.6% authority bias while criminal justice shows only 2.8%; (3) structured decomposition, where the LLM extracts features and a deterministic rubric decides, reduces flip rates by up to 100% (median 49% across 9 models). We demonstrate an ICE-guided detect-diagnose-mitigate-verify loop achieving cumulative 78% bias reduction via iterative prompt patching. Validation against real COMPAS recidivism data shows COMPAS-derived flip rates exceed pooled synthetic rates, suggesting our benchmark provides a conservative estimate of real-world bias. Code and data are publicly available.

Related papers

• Beyond Aggregate Calibration: Decomposing Income-Conditional Recall Disparities in Automated Credit Default Prediction [0.0]
  High-income defaulters are disproportionately classified as label noise relative to low-income defaulters.<n>A 16.86 percentage point gap in true positive rate (recall) between high- and low-income borrowers who ultimately defaulted.
  arXiv  Detail & Related papers  (2026-08-08T15:57:15Z)
• Preserving Disagreement: Architectural Heterogeneity and Coherence Validation in Multi-Agent Policy Simulation [0.0]
  Multi-agent deliberation systems using large language models (LLMs) are increasingly proposed for policy simulation.<n>They suffer from artificial consensus: evaluator agents converge on the same option regardless of their assigned value perspectives.<n>We present the AI Council, a three-phase deliberation framework, and conduct 120 deliberations across two policy scenarios to test two interventions.
  arXiv  Detail & Related papers  (2026-04-29T11:47:28Z)
• CogBias: Measuring and Mitigating Cognitive Bias in Large Language Models [8.81733256907103]
  Large Language Models (LLMs) are increasingly deployed in high-stakes decision-making contexts.<n>We show that cognitive biases emerge systematically across all four families of cognitive biases: Judgment, Information Processing, Social, and Response.<n>We apply activation steering to modulate biased behavior, achieving 26--32% reduction in bias score.
  arXiv  Detail & Related papers  (2026-04-01T20:22:14Z)
• HALF: Harm-Aware LLM Fairness Evaluation Aligned with Deployment [52.374772443536045]
  HALF (Harm-Aware LLM Fairness) is a framework that assesses model bias in realistic applications and weighs the outcomes by harm severity.<n>We show that HALF exposes a clear gap between previous benchmarking success and deployment readiness.
  arXiv  Detail & Related papers  (2025-10-14T07:13:26Z)
• The Bias is in the Details: An Assessment of Cognitive Bias in LLMs [1.9651471153360769]
  Large Language Models (LLMs) are increasingly embedded in real-world decision-making processes.<n>This paper presents a large-scale evaluation of eight well-established cognitive biases across 45 LLMs.<n>We find that both model size and prompt specificity play a significant role on bias susceptibility.
  arXiv  Detail & Related papers  (2025-09-26T19:09:42Z)
• TrustJudge: Inconsistencies of LLM-as-a-Judge and How to Alleviate Them [58.04324690859212]
  Large Language Models (LLMs) as automated evaluators (LLM-as-a-judge) has revealed critical inconsistencies in current evaluation frameworks.<n>We identify two fundamental types of inconsistencies: Score-Comparison Inconsistency and Pairwise Transitivity Inconsistency.<n>We propose TrustJudge, a probabilistic framework that addresses these limitations through two key innovations.
  arXiv  Detail & Related papers  (2025-09-25T13:04:29Z)
• GANDiff FR: Hybrid GAN Diffusion Synthesis for Causal Bias Attribution in Face Recognition [0.0]
  We introduce GANDiff FR, the first synthetic framework that precisely controls demographic and environmental factors to measure, explain, and reduce bias with reproducible rigor.<n>We synthesize 10,000 demographically balanced faces across five cohorts validated for realism via automated detection and human review.<n> Benchmarking ArcFace, CosFace, and AdaFace under matched operating points shows AdaFace reduces inter-group TPR disparity by 60%.<n>Despite around 20% computational overhead relative to pure GANs, GANDiff FR yields three times more attribute-conditioned variants.
  arXiv  Detail & Related papers  (2025-08-15T09:05:57Z)
• Mitigating Subgroup Disparities in Multi-Label Speech Emotion Recognition: A Pseudo-Labeling and Unsupervised Learning Approach [53.824673312331626]
  Implicit Demography Inference (IDI) module uses k-means clustering to mitigate bias in Speech Emotion Recognition (SER)<n>Experiments show that pseudo-labeling IDI reduces subgroup disparities, improving fairness metrics by over 28%.<n>Unsupervised IDI yields more than a 4.6% improvement in fairness metrics with a drop of less than 3.6% in SER performance.
  arXiv  Detail & Related papers  (2025-05-20T14:50:44Z)
• Assessing Judging Bias in Large Reasoning Models: An Empirical Study [99.86300466350013]
  Large Reasoning Models (LRMs) like DeepSeek-R1 and OpenAI-o1 have demonstrated remarkable reasoning capabilities.<n>We present a benchmark comparing judging biases between LLMs and LRMs across both subjective preference-alignment datasets and objective fact-based datasets.
  arXiv  Detail & Related papers  (2025-04-14T07:14:27Z)
• Exploring Response Uncertainty in MLLMs: An Empirical Evaluation under Misleading Scenarios [49.53589774730807]
  Multimodal large language models (MLLMs) have recently achieved state-of-the-art performance on tasks ranging from visual question answering to video understanding.<n>We reveal a response uncertainty phenomenon: twelve state-of-the-art open-source MLLMs overturn a previously correct answer in 65% of cases after receiving a single deceptive cue.
  arXiv  Detail & Related papers  (2024-11-05T01:11:28Z)
• Identifying and Mitigating Social Bias Knowledge in Language Models [52.52955281662332]
  We propose a novel debiasing approach, Fairness Stamp (FAST), which enables fine-grained calibration of individual social biases.<n>FAST surpasses state-of-the-art baselines with superior debiasing performance.<n>This highlights the potential of fine-grained debiasing strategies to achieve fairness in large language models.
  arXiv  Detail & Related papers  (2024-08-07T17:14:58Z)
• Fairness-enhancing mixed effects deep learning improves fairness on in- and out-of-distribution clustered (non-iid) data [6.596656267996196]
  We propose the Fair Mixed Effects Deep Learning (Fair MEDL) framework.<n>This framework quantifies cluster-invariant fixed effects (FE) and cluster-specific random effects (RE) through: 1) a cluster adversary for learning invariant FE, 2) a Bayesian neural network for RE, and 3) a mixing function combining FE and RE for final predictions.<n>Fair MEDL framework improves fairness by 86.4% for Age, 64.9% for Race, 57.8% for Sex, and 36.2% for Marital status, while maintaining robust predictive performance.
  arXiv  Detail & Related papers  (2023-10-04T20:18:45Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.