[Medical ML] Physician Resistance: Causes and Solutions

Posted on by Yoman

Physician Resistance: Causes and Solutions

Article #12 in Medical ML for Ukrainian Doctors Series

By Oleh Ivchenko | Researcher, ONPU | Stabilarity Hub | February 8, 2026

📋 Key Questions Addressed

  1. What psychological, professional, and structural factors drive physician resistance to medical AI?
  2. How does familiarity with AI influence physician attitudes?
  3. What evidence-based approaches successfully transform resistance into engagement?

Context: Why This Matters for Ukrainian Healthcare

Understanding physician resistance isn’t optional—it’s essential. Globally, despite over $66 billion invested in healthcare AI, adoption remains stubbornly low. For ScanLab and Ukrainian healthcare modernization, converting physician skepticism into informed engagement will determine success.

The Resistance Spectrum: From Skepticism to Fear

“`mermaid
graph LR
A[Skepticism
Mild] –> B[Reluctance
B –> C[Anxiety
C –> D[Resistance
D –> E[Fear
“`

Attitude Description Intensity
Skepticism Questioning stance, demands evidence ● Mild
Reluctance Hesitation and unwillingness ● Moderate
Anxiety Emotional concerns about risks ● Elevated
Resistance Active, deliberate opposition ● High
Fear Intense emotional response, avoidance ● Severe
💡 Key Insight: These attitudes are interconnected with feedback loops. Unaddressed skepticism deepens into anxiety; prolonged resistance reinforces fear. Early intervention at the skepticism stage prevents escalation.

The Root Causes: Intrinsic and Extrinsic Factors

Intrinsic Factors (Professional Identity)

Factor Description Prevalence
Professional autonomy threat Fear of losing control over clinical decisions 67%
Deskilling concerns Worry that AI will erode clinical expertise 54%
Job displacement Fear of replacement by AI systems 48%
Competence questions Concern about inability to evaluate AI recommendations 41%

Extrinsic Factors (Patient Care & Systems)

Factor Description Prevalence
Liability uncertainty Unclear who is responsible when AI errs 63%
Patient relationship impact Fear AI will depersonalize care 52%
Black box opacity Cannot explain AI reasoning to patients 47%
Workflow disruption Concern about added complexity and time 31%

The Liability Paradox: Damned If You Do, Damned If You Don’t

⚖️ The Dilemma

Follow AI (AI wrong) → Potential liability for blind algorithmic following
Override AI (physician wrong) → Potential liability for ignoring decision support
Fail to use AI → Future liability as AI becomes standard of care

“IT staff reported being asked by worried physicians about what would happen if they diverged from the CDSS recommendation (and struggled to answer, as the legal framework is unclear).”

— Oxford Medical Law Review, 2023

The Familiarity Factor: Experience Transforms Attitudes

A landmark 2025 JMIR study (498 physicians) revealed the most important finding:

+91%

Higher enthusiasm (familiar vs. unfamiliar)

+59%

Lower skepticism (familiar vs. unfamiliar)

🔑 Critical Finding: Age and medical specialty had NO significant influence on attitudes. Experience with AI—not demographics—determines acceptance.

What Works: Evidence-Based Solutions

1. Early Physician Engagement

Phase Physician Role
Needs Assessment Identify actual clinical pain points
Vendor Evaluation Assess clinical utility claims
Pilot Design Design realistic testing protocols
Implementation Champion adoption among peers
Monitoring Report real-world performance issues

2. Prioritize Explainable AI

✅ Explainable AI

  • Physician can see why AI flagged finding
  • Can challenge basis for decisions
  • Higher liability comfort
  • Enables learning, not just following

❌ Black Box AI

  • Cannot review reasoning
  • Blind acceptance or rejection
  • Lower trust
  • Harder to explain to patients

3. Address the Psychological Progression

Current State Intervention Strategy
Skepticism Provide evidence, address specific concerns
Reluctance Offer low-stakes exposure, peer testimonials
Anxiety Psychological support, clear liability guidance
Resistance One-on-one engagement, address specific grievances
Fear May require organizational culture change

The Chief Physician Effect: Leadership Matters

📊 Unexpected Finding

Chief physicians showed significantly lower skepticism than residents (p=.01)

Strategic Implication: Engage chief physicians as AI champions. Their endorsement carries weight with junior staff.

Conclusions

✅ Experience > Demographics

Familiarity with AI predicts acceptance; age and specialty do not

🎓 Early Engagement

Involve physicians from selection through monitoring

⚖️ Clarify Liability

Undefined liability creates a chilling effect on adoption

🔍 Explainability Matters

Prioritize tools where physicians can see reasoning

Questions Answered

✅ What drives physician resistance?

Professional autonomy threat (67%), liability uncertainty (63%), deskilling concerns (54%), and patient relationship impacts (52%).

✅ How does familiarity influence attitudes?

Physicians familiar with AI show 91% higher enthusiasm and 59% lower skepticism. Age and specialty have no significant effect.

✅ What approaches work?

Early engagement, explainable AI tools, hands-on training, addressing the psychological progression, and engaging senior physicians as champions.

Next in Series: Article #13 – The 2007-2012 Golden Age (Ancient IT)

Series: Medical ML for Ukrainian Doctors | Stabilarity Hub Research Initiative

Author: Oleh Ivchenko | ONPU Researcher | Stabilarity Hub