Skip to content

R0043/2026-04-01/Q001/H2

Research R0043 — Sycophancy Vocabulary
Run 2026-04-01
Query Q001
Hypothesis H2

Statement

Some domains have not developed specific terminology for this phenomenon and either use generic terms (e.g., "bias," "accuracy") or have no vocabulary at all for the concept.

Status

Current: Partially Supported

Supporting Evidence

Evidence Summary
SRC07-E01 Enterprise software evaluation and UX/product design lack domain-specific terms, using general concepts like "hallucination rate" and "satisfaction-accuracy tradeoff"
SRC09-E01 Article explicitly identifies that regulated industries describe manifestations without standardized nomenclature

Contradicting Evidence

Evidence Summary
SRC03-E01 Defense and aviation have well-established terminology dating to early 2000s
SRC05-E01 Human-AI interaction research has a rich vocabulary with formal definitions

Reasoning

This hypothesis is partially supported. While most domains have some terminology, the depth and specificity varies enormously. Aviation and defense have decades-old, well-defined terminology. AI safety has specific recent terminology. But enterprise software evaluation, academic integrity, and UX/product design largely borrow terms from other fields rather than developing domain-native vocabulary for this specific phenomenon.

Relationship to Other Hypotheses

H2 is a refinement of H1 — it holds for 2-3 of the 8 domains (academic integrity, enterprise software, UX) while H1 holds for the remainder. H2 and H1 are compatible rather than contradictory.