Rater State Bias in RLHF Preference Data: An Audit Framework

arXiv:2607.16195 · cs.AI · Submitted 2026-04-14 · Read on arXiv

cs.AI

Submitted: 2026-04-14

Updated: 2026-09-28

Terminology

• Aggregation — 2× in this paper · explained in 1 episode
• Reinforcement Learning — 1× in this paper · explained in 4 episodes

Sources

• Discovering Language Model Behaviors with Model-Written Evaluations
• Towards Understanding Sycophancy in Language Models
• How RLHF Amplifies Sycophancy
• Scaling Laws for Reward Model Overoptimization in Direct Alignment Algorithms
• A Systematic Study of Bias Amplification
• An Effective Theory of Bias Amplification
• Deep Learning is Robust to Massive Label Noise
• The GPT-4o Shock Emotional Attachment to AI Models and Its Impact on Regulatory Acceptance: A Cross-Cultural Analysis of the Immediate Transition from GPT-4o to GPT-5
• How AI and Human Behaviors Shape Psychosocial Effects of Extended Chatbot Use: A Longitudinal Randomized Controlled Study
• Investigating Affective Use and Emotional Well-being on ChatGPT

Related papers

• MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
• Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
• Agents' Last Exam
• The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
• MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
• Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems