LLM Digest

Story

arxiv_llm_reliability · Oct 22, 2026 · paper

Source brief

Who Is Sycophantic to Whom? Confident LLMs, Human Deference, and the Collapse of Trust Calibration

arxiv.orgOct 22, 2026
original source linked

In brief

Research on sycophancy asks whether large language models (LLMs) bend towards what their users want to hear. We ask the reverse question: when an LLM is confident, consistent, and wrong, do users bend towards it? In a...

Feed lens

evaluation

Read the original at arxiv.org →Open in live feed Read that day’s brief

Who Is Sycophantic to Whom? Confident LLMs, Human Deference, and the Collapse of Trust Calibration

Earlier in this thread 4 items

When LLMs Forget What They Believe: Characterizing State Integrity Collapse in Long-Horizon Dialogues

S-EDL: Eliciting Self-Evidence from Sequence Likelihoods for Semantic Calibration of LLMs

Who judges the judge? Three helpful ways to audit your LLM evaluators

Robot Learning from Human Demonstrations: Handwritten Alphabet Trajectories and Human-Likeness Evaluation