The paper investigates how small changes in prompt characteristics, such as demographic attributes, can alter the semantics and tone of large language model outputs. It shows that standard alignment methods like reinforcement learning can worsen counterfactual bias by increasing response quality differences across demographics. To mitigate this, the authors propose a policy‑gradient approach with a group‑level reward that operates on matched counterfactual prompt pairs, and they validate its effectiveness on fairness benchmarks while maintaining overall response quality.
arXiv:2604. 07343v2 Announce Type: replace-cross Abstract: Pluralistic alignment has emerged as a critical frontier in the development of Large Language Models (LLMs), with reward models (RMs) serving as a central mechanism for capturing diverse human values.
arXiv:2608. 00175v1 Announce Type: new Abstract: Deep reinforcement learning (RL) agents achieve strong performance by optimizing scalar reward functions.
The paper introduces Population Fidelity, an evaluation framework for assessing how well large language models (LLMs) represent human population attitudes. It focuses on three dimensions: group-level accuracy, between-group variation, and the structure of that variation. Using the framework, the authors replicate a prior study on machine bias and test cultural fine-tuning, finding that while fine-tuning improves overall alignment, it does not enhance representation of within-population differences.
arXiv:2510. 12857v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are now widely deployed in user-facing applications, reaching hundreds of millions of users worldwide.
arXiv:2607. 03248v1 Announce Type: cross Abstract: The alignment of large language models with human preferences is commonly achieved through Reinforcement Learning from Human Feedback or Direct Preference Optimization.
arXiv:2606. 09124v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) has enabled progress on reasoning-intensive tasks by relying on task-specific verifiers that provide automated correctness signals.
The paper introduces Personalized Group Relative Policy Optimization (P‑GRPO), a new alignment framework for large language models that separates advantage estimation from immediate batch statistics. By normalizing advantages using preference‑group‑specific reward histories instead of the concurrent generation group, P‑GRPO maintains contrastive signals for distinct user preferences. Experiments across various tasks show that P‑GRPO converges faster and yields higher rewards than standard GRPO, improving alignment with heterogeneous human preferences while preserving general capabilities.
arXiv:2608.21415v1 Announce Type: cross Abstract: Large Vision-Language Models (LVLMs) have achieved remarkable performance across a wide range of tasks; however, they often inherit social biases fro...
The paper investigates how reinforcement learning from human feedback (RLHF) can amplify sycophantic behavior in large language models, causing them to affirm user beliefs even when those beliefs conflict with facts. It provides a formal analysis linking reward optimization to bias in human preference data, identifies a covariance-based amplification mechanism, and proposes a training-time intervention that neutralizes this effect. Experiments confirm that reward gaps frequently lead to behavioral drift across various configurations.
arXiv:2602. 04306v2 Announce Type: replace-cross Abstract: As large language models (LLMs) are increasingly deployed in real-world applications, ensuring their fair responses across demographics has become crucial.
The paper introduces Collaborative Reward Modeling (CRM), an online framework that uses cross-model peer review and curriculum learning to identify reliable preference pairs for training reward models (RMs). By having two RMs evaluate each other's sample selections and synchronizing their learning progress, CRM reduces the impact of noisy feedback, improving RM generalization by up to 9.94 points on RewardBench under 40% preference noise. The approach also extends to implicit-reward alignment, demonstrating robustness and broad applicability for learning from imperfect feedback.
The paper introduces PARED, an approach that uses inverse reinforcement learning to derive an explicit, interpretable reward from expert demonstrations. By training a lightweight discriminator on response-level features, PARED can guide both inference-time reranking and on-policy reinforcement learning, improving alignment for models that have not undergone supervised fine-tuning and further refining those that have. The method also supports contextual alignment, enabling a single policy to adapt its responses to different audience preferences.