The vendor report says the tool is adopted. If you stand at the desk for a morning, you may see something else.
For the first week, people try it. Then, without a meeting or a formal complaint, they go back to the pathway they already trust. The license is still paid for. The dashboard still looks busy. On the floor, the tool is quietly unused.
I spent about twenty years in the emergency department, and I have spent the years since talking with clinicians and health system leaders about where AI belongs in care. The pattern I hear most often is not a loud refusal. It is a quiet walk-away.
Leaders often call this resistance. I do not think that word is precise enough. Most clinicians I know will use a tool that helps them take better care of the person in front of them. They will also stop using a tool that asks them to trust something they cannot explain, cannot challenge, or cannot see working in a setting like theirs.
Real physician voices, twice a week
Free, and one click to unsubscribe.
The silence is not always stubbornness. Sometimes it is judgment. The cost of speaking up can feel higher than the cost of clicking the old pathway.
I have been applying Minda Harts’ Seven Trust Languages to this problem. The framework is hers, from her book Talk to Me Nice: The Seven Trust Languages for a Better Workplace. It was written as a language for trust at work, not as a clinical protocol. I am not offering it as a validated checklist, a scoring system, or a substitute for evidence. It is a useful lens for a question I keep hearing: Why did a capable team stop using a tool that was supposed to help them?
Sensitivity
A clinician may stop using a tool when it treats a person as a score and little else. Picture a risk score that keeps treating a familiar patient as average because it cannot hold the details a nurse has watched accumulate over several hours. The score may be tidy. The person in the chair is not. If the tool has no room for that, trust erodes in small moments, and use fades with it.
Transparency
People also walk away when they cannot see why the model leaned one way. If I cannot explain a flag to a colleague or a family in ordinary language, I am being asked to put my name on a black box. In medicine, that is not a small ask. A tool that cannot show its work, even in plain words, invites quiet nonuse.
Security
Some clinicians stop reporting problems long before they stop using the tool, and then the second follows the first. If raising a concern feels as if it could put a job, a license, or a patient’s care at risk, the safe move is to work around the system. A safety tool that makes the person who flags it feel exposed will not stay in use for long.
Demonstration
A successful presentation is not the same as proof in a setting like ours. I have sat through talks that looked excellent in a quiet conference room. An emergency department is not a quiet conference room. If no one can show that the tool has been tried with our volume, our interruptions, and our mix of patients, a careful clinician will keep it at arm’s length.
Feedback
When the tool is wrong, there has to be a real way to say so, and someone has to listen. If the only option is a form that disappears, or a button that does nothing visible, clinicians learn that correction is theater. They will not keep feeding a system that does not learn from them.
Acknowledgment
Trust also fades when a person catches what the model missed and the credit goes to the software, or to no one. The nurse who notices the miss, the resident who questions a drafted note, the patient who says the summary is wrong: If those catches are treated as inconveniences, people stop offering them. A tool that cannot share credit will not keep human partners.
Follow-through
The last language is the one I see skipped most often. After the kickoff, do training, oversight, and human review actually show up on the floor? A tool launched with a celebration and then left without owners, refreshers, or a place to escalate problems is a tool that will be abandoned in weeks. Clinicians notice what arrives after the applause.
Put together, these seven questions explain a lot of “low adoption” without blaming the people closest to the patient. The team may not be resisting AI. They may be answering, in their own quiet way, a trust question that was never asked out loud.
If a tool in your shop has gone unused, it may help to ask which language failed. Was the person reduced to a score? Was the reasoning invisible? Was it unsafe to speak up? Was the proof only a pitch? Was there no real way to correct the tool? Was credit withheld? Did the follow-through never arrive?
You do not need a new committee to start. You need an honest conversation with the people who were supposed to use the tool, and a willingness to hear that the problem may not be their attitude.
This is a reflection for clinicians and leaders, not medical advice and not a protocol. Minda Harts gave workplaces a language for trust. In a care setting, that language can also explain why a clinician closes a tab and does not open it again.
Harvey Castro, known as DR GPT™, is an emergency physician, author, keynote speaker, and physician futurist. He is advisor to the CEO of Phantom Space. He writes and speaks about artificial intelligence in health care, digital health, space medicine, and the future of human-centered care. Explore his speaking topics and availability.
He is the author of Bing Copilot and Other LLM: Revolutionizing Healthcare With AI, Solving Infamous Cases with Artificial Intelligence, The AI-Driven Entrepreneur: Unlocking Entrepreneurial Success with Artificial Intelligence Strategies and Insights, ChatGPT and Healthcare: The Key To The New Future of Medicine, ChatGPT and Healthcare: Unlocking The Potential Of Patient Empowerment, Revolutionize Your Health and Fitness with ChatGPT’s Modern Weight Loss Hacks, Success Reinvention, and Apple Vision Healthcare Pioneers: A Community for Professionals & Patients.
He shares updates on LinkedIn, YouTube, Facebook, Instagram, and X. #DRGPT
