R0043/2026-03-28/Q001/SRC09/E01¶
Healthcare acquiescence problem and related terminology
URL: https://www.mdpi.com/2076-3417/16/2/710
Extract¶
Healthcare-specific terminology: - Acquiescence problem: "Current AI systems passively confirm rather than challenge clinicians' hypotheses, reinforcing cognitive biases such as anchoring and premature closure" - Automation bias: Used extensively in healthcare CDSS literature for inappropriate reliance on AI recommendations - Alert fatigue: Clinician desensitization from excessive automated alerts, leading to inappropriate override of ALL alerts - Deskilling: Loss of clinical diagnostic reasoning abilities due to overreliance on AI tools - Commission errors: Following incorrect automated advice - Omission errors: Failing to notice when automation does not flag a problem
The Dialogic Reasoning Framework proposes three roles to counter acquiescence: 1. Framework Coach: Guides structured reasoning 2. Socratic Guide: Asks probing questions 3. Red Team Partner: Presents evidence-based alternatives
JUDGMENT: Healthcare has the richest vocabulary for the phenomenon after AI safety. The "acquiescence problem" is the closest domain-specific equivalent to "sycophancy" — it describes system behavior (passive confirmation) rather than just human behavior (overreliance). However, it frames the AI as passive (failing to challenge) rather than active (deliberately agreeing), which is a meaningful distinction from sycophancy.
Relevance to Hypotheses¶
| Hypothesis | Relationship | Strength |
|---|---|---|
| H1 | Supports | Healthcare has multiple domain-specific terms |
| H2 | Contradicts | Rich vocabulary exists |
| H3 | Supports | "Acquiescence problem" is the closest to system-side framing but still treats it as passive failure rather than active behavior — maintaining the asymmetry thesis |
Context¶
The healthcare context is particularly important because patient safety makes the stakes of sycophancy-adjacent behavior concrete and measurable. The 26% increase in error rates from automation bias (Parasuraman & Manzey) demonstrates that the vocabulary gap has real consequences.