Archive · 20 Sep 2026
Doctors Resist Medical AI Beyond Scans Over Limited Evidence
AI already helps some doctors read scans and spot abnormalities.
Doctors are more cautious about AI that suggests treatments or writes clinical decisions.
They say these tools have not been tested enough in real hospitals.
A 2026 survey of US doctors and nurses found 74% worried AI could weaken clinicians' skills.
Another 74% worried AI could make up false information and sound confident.
These problems could be dangerous if people stop checking the computer carefully.
Supporters say AI might reduce paperwork, find drug interactions, and give clinicians more time with patients.
The article says hospitals should validate tools in real life and create rules before using them widely.
Clinicians are resisting AI tools that go beyond scan reading and abnormality detection.
Doctors say broader tools lack enough real-world evidence to prove they work safely.
A 2026 survey found 74% of US doctors and nurses feared AI could cause deskilling.
Another 74% worried that AI could confidently invent false information, including fake medical studies.
The article says real-world validation and governance could help hospitals distinguish useful tools from unproven ones.
- Who
- Doctors, nurses, hospitals, and vendors developing medical AI tools.
- What
- Clinicians are resisting broader medical AI adoption because they say the tools lack sufficient real-world evidence.
- When
- A 2026 survey measured doctors' and nurses' concerns; the article does not provide a more specific date for the wider debate.
- Where
- The survey covered doctors and nurses in the United States.
- Why
- Broader AI tools can be difficult to validate, and clinicians fear deskilling and confident false information.
Evidence-first clinicians
Potential AI adopters
Proof before adoption
Evidence-first clinicians
Clinicians say vendors must provide real-world evidence that tools work before hospitals rely on them for treatment suggestions or clinical decisions.
Potential AI adopters
Supporters argue that excessive caution could delay useful tools and impose costs on patients.
Reliability risks
Evidence-first clinicians
Doctors and nurses fear AI may weaken professional skills and generate convincing false information, making errors harder to detect.
Potential AI adopters
AI may still help clinicians by reducing administrative work, identifying drug interactions, and allowing more patient contact.
Implementation approach
Evidence-first clinicians
Hospitals should use validation and governance to ensure tools are safe and effective.
Potential AI adopters
The article does not describe a case for unrestricted adoption, but suggests useful tools can earn trust through evidence and oversight.
- Established AI use
- AI has a foothold in reading scans and flagging abnormalities.
- Survey year
- 2026
- Survey population
- US doctors and nurses
- Deskilling concern
- 74% worried that relying on AI could erode clinicians' abilities.
- Hallucination concern
- 74% worried that AI could invent incorrect information and present it confidently.
- Potential benefits
- AI could reduce paperwork, catch drug interactions, and free clinicians for patient contact.
- Suggested path forward
- Hospitals should conduct real-world validation and establish governance before wider adoption.