Research preview. Independent expert review is pending. The worked analyses are hypothetical. Method and corrections
AIAIGovSpec
Source register

scholarship

Towards Understanding Sycophancy in Language Models

ICLR 2024 · published 2023-10-20

Open the source

sharma-2024-sycophancy · last checked 2026-10-18 · voice: industry · origin: global north

Developer research: the paper states that all authors were at Anthropic, a developer of AI assistants (and of the model that assisted this edit). Published at ICLR 2024; arXiv version cited.

Claims that cite this source · 1

  1. Source assertion· confidence medium· passage checked

    Research by an AI developer found that leading assistants behave sycophantically and that human preference data used in training tends to reward answers that match users’ views.

    Used in: Sycophancy and engagement-driven agreement