Skip to content

R0056/2026-04-01/C003/SRC01/E01

Research R0056 — RLHF Yes-Men Claims v2
Run 2026-04-01
Claim C003
Source SRC01
Evidence SRC01-E01
Type Reported

Primary evidence for C003

URL: See source scorecard

Extract

Multiple papers demonstrate sycophancy amplification originates from systematic bias in preference data, not algorithmic failures in RLHF itself.

Relevance to Hypotheses

Hypothesis Relationship Strength
H1 Supports See assessment
H2 Supports See assessment
H3 Contradicts See assessment

Context

See assessment.md for full context.