E01¶


Research	R0043 — Sycophancy Vocabulary
Run	2026-04-01
Query	Q001
Source	SRC08
Evidence	SRC08-E01
Type	Analytical

Aviation domain terminology and its acknowledged inadequacy for AI-era interactions

URL: https://www.mdpi.com/2673-7590/5/2/42

Extract¶

Aviation human factors vocabulary:

Automation complacency — first used in connection with aviation accidents where pilots/controllers failed to check systems sufficiently. Defined by NASA ASRS as "self-satisfaction that may result in non-vigilance based on an unjustified assumption of satisfactory system state."
Overtrust — inaccurate calibration where human trust exceeds machine capability
Undertrust — the opposite; human trust falls below machine capability
Four-category taxonomy (Parasuraman & Riley, 1997):
Use — appropriate reliance on automation
Misuse — over-reliance
Disuse — under-reliance / disengagement
Abuse — poor allocation decisions between human and machine

Critical finding: "existing blanket terms such as complacency and over-trust with respect to automation are probably not nuanced enough to capture the full extent of transactional relationships that will exist between human crews and AI support systems." The authors argue that as AI systems become more advanced, taxonomies need updating with new terms.

JUDGMENT: Aviation has the most mature vocabulary for this domain but explicitly acknowledges it is insufficient for AI-era interactions. This is a key finding — even the domain with the longest history of addressing this problem recognizes the vocabulary gap.

Relevance to Hypotheses¶

Hypothesis	Relationship	Strength
H1	Supports	Confirms aviation has specific, long-established terminology
H2	N/A	Aviation has vocabulary; it's other domains that lack it
H3	Strongly supports	Aviation researchers themselves say their terms may not map to AI interactions

Context¶

The Parasuraman & Manzey (2010) paper on "Complacency and Bias in Human Use of Automation" is the foundational text for the use/misuse/disuse/abuse taxonomy. The 2025 paper extends this by acknowledging the taxonomy's limitations for AI.