Skip to content

R0043/2026-04-01/Q001

Query: What terms do different industries and disciplines use to describe AI behavior that prioritizes user agreement, comfort, or satisfaction over accuracy, correctness, or safety? Map the complete vocabulary across: AI safety research, defense/military AI, healthcare AI, financial services AI, aviation/FAA, academic integrity, enterprise software evaluation, and UX/product design. For each domain, identify the specific terms used for the phenomenon that AI researchers call "sycophancy."

BLUF: The phenomenon AI researchers call "sycophancy" maps to a rich but fragmented vocabulary across eight domains. No two fields use the same primary term. The terms describe different facets of the same system: sycophancy names the model behavior, automation bias names the human cognitive response, overreliance names the behavioral outcome, and domain-specific terms name the phenomenon as it manifests in each context. A comprehensive map is constructable but reveals related concepts with different causal framings, not simple synonyms.

Probability: N/A (open-ended query) | Confidence: Medium


Summary

Entity Description
Query Definition Query text, scope, status
Assessment Full analytical product with reasoning chain and cross-domain vocabulary map
ACH Matrix Evidence x hypotheses diagnosticity analysis
Self-Audit ROBIS-adapted 5-domain audit (process + source verification)

Hypotheses

ID Hypothesis Status
H1 Each domain has specific terminology; comprehensive map constructable Supported
H2 Some domains lack specific terminology, using generic or borrowed terms Partially Supported
H3 Terms describe fundamentally different phenomena, not the same thing under different names Partially Supported

Searches

ID Target Results Selected
S01 AI safety sycophancy terminology 10 4
S02 Defense/military automation bias terminology 10 2
S03 Healthcare acquiescence/deference terminology 10 2
S04 Financial, aviation, UX terminology 30 3
S05 Cross-cutting overreliance/acquiescence terms 20 2

Sources

Source Description Reliability Relevance
SRC01 NN/g sycophancy behavioral categories High High
SRC02 TechPolicy.Press regressive/progressive taxonomy Medium-High High
SRC03 CSET defense automation bias brief High High
SRC04 PMC cross-domain vocabulary mapping High High
SRC05 Oxford/Cambridge overreliance taxonomy High High
SRC06 Braun acquiescence bias counter-finding Medium-High Medium
SRC07 IEEE Spectrum engineering perspective High Medium-High
SRC08 Aviation human factors terminology Medium-High High
SRC09 Cross-domain vocabulary gap analysis Medium High

Cross-Domain Vocabulary Map

Domain Primary Terms Causal Focus
AI Safety Sycophancy, reward hacking, sandbagging Model behavior
Defense/Military Automation bias, complacency, overtrust, miscalibrated trust Human cognition
Healthcare Acquiescence, deference, acquiescence problem Clinical decision quality
Aviation/FAA Automation complacency, overtrust/undertrust, use/misuse/disuse/abuse Human factors
Financial Services Model risk, effective challenge, challenge function Governance mechanism
Academic Integrity Confirmation bias amplification (borrowed) Detection
Enterprise Software Hallucination rate, accuracy metrics (borrowed) Measurement
UX/Product Design Satisfaction-accuracy tradeoff, dark patterns (borrowed) Design tension
Cross-cutting Overreliance, appropriate reliance, calibrated trust Human behavior

Revisit Triggers

  • Publication of a formal cross-domain taxonomy or glossary by NIST, ISO, or IEEE that includes sycophancy-related terms
  • The EU AI Act implementing rules or guidance that name sycophancy or adopt domain-specific alternatives
  • DOD or FAA issuing updated human-AI teaming terminology that explicitly addresses LLM-era sycophancy
  • Major AI incident in a regulated industry where the vocabulary gap is cited as a contributing factor