ReCrit frames critic interaction as a correctness-transition problem and uses quadrant-based RL rewards to improve LLM performance on scientific reasoning benchmarks by rewarding corrections and robustness while penalizing sycophancy.
Beacon: Single-turn diagnosis and mitigation of latent sycophancy in large language models
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
abstract
Large language models internalize a structural trade-off between truthfulness and obsequious flattery, emerging from reward optimization that conflates helpfulness with polite submission. This latent bias, known as sycophancy, manifests as a preference for user agreement over principled reasoning. We introduce Beacon, a single-turn forced-choice benchmark that isolates this bias independent of conversational context, enabling precise measurement of the tension between factual accuracy and submissive bias. Evaluations across twelve state-of-the-art models reveal that sycophancy decomposes into stable linguistic and affective sub-biases, each scaling with model capacity. We further propose prompt-level and activation-level interventions that modulate these biases in opposing directions, exposing the internal geometry of alignment as a dynamic manifold between truthfulness and socially compliant judgment. Beacon reframes sycophancy as a measurable form of normative misgeneralization, providing a reproducible foundation for studying and mitigating alignment drift in large-scale generative systems.
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 2years
2026 2verdicts
UNVERDICTED 2roles
background 1polarities
background 1representative citing papers
The paper introduces the Proxy Compression Hypothesis as a unifying framework explaining reward hacking in RLHF as an emergent result of compressing high-dimensional human objectives into proxy reward signals under optimization pressure.
citing papers explorer
-
ReCrit: Transition-Aware Reinforcement Learning for Scientific Critic Reasoning
ReCrit frames critic interaction as a correctness-transition problem and uses quadrant-based RL rewards to improve LLM performance on scientific reasoning benchmarks by rewarding corrections and robustness while penalizing sycophancy.
-
Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges
The paper introduces the Proxy Compression Hypothesis as a unifying framework explaining reward hacking in RLHF as an emergent result of compressing high-dimensional human objectives into proxy reward signals under optimization pressure.