Beyond the Scribe: What AI Could Actually Become in Healthcare
RUSSEL MCKENNA, D.O., pediatrician, Canyon Medical Group
We are living through one of the most important technological shifts in modern history, yet much of the conversation around artificial intelligence in healthcare feels surprisingly small.
Most discussions focus on efficiency: Can AI write notes? Can it summarize visits? Can it reduce burnout? Those are useful questions. But they are also limited ones.
As a physician, I think we may be underestimating what AI could actually become.
Right now, many doctors interact with AI the same way many patients do: They ask it questions and get answers. In medicine, most systems are primarily using AI as a documentation tool or digital scribe.
But what if the real value of AI is not in answering questions?
What if its greatest value is in helping us ask better ones?
Medicine has always depended on thoughtful history-taking. Long before advanced imaging, genetics, or artificial intelligence, physicians relied on asking good questions and recognizing patterns hidden inside a patient’s story.
The challenge is that no physician can personally experience every disease, every atypical presentation, or every uncommon manifestation of a common condition. Medicine is simply too large.
Doctors provide care based on their knowledge and experience. But every doctor also has blind spots created by the natural limits of individual training and exposure. That is not failure. That is reality.
This is where AI becomes interesting.
Not because it replaces doctors. Not because it independently diagnoses patients. But because it can help patients and healthcare teams explore concerns more thoroughly before important details are overlooked.
Imagine a patient with fatigue, headaches, abdominal pain, or weight changes. In a rushed healthcare system, there may only be enough time to touch the surface. But an AI-guided intake process could help explore medications, travel, family history, exposures, mental health overlap, disease progression, sleep patterns, and subtle clues the patient may not realize are important.
The goal is not for AI to diagnose the patient. The goal is to help the physician begin the encounter with a broader, more organized clinical picture.
One experience that reinforced my thinking involved a 3-year-old child who initially presented to the emergency room with what appeared to be a relatively unimpressive rash in his diaper.
The child was diagnosed with a yeast infection and treated accordingly, but the treatment did not help. Two days later, the family presented to our pediatric clinic.
What changed the encounter was not that AI “diagnosed” the patient. It did not. What mattered was that a more thoughtful and structured questioning process helped develop the story more completely.
Additional details surfaced that had not previously been connected: the child had experienced a sore throat roughly two weeks earlier, both parents had recently been diagnosed with strep throat, and the family had been applying topical steroid cream to the rash, altering its appearance and making it appear less inflamed.
Those details mattered.
The child ultimately tested positive for Group A Streptococcus and improved quickly once appropriate treatment was started.
In pediatrics, atypical presentations are common. Perianal streptococcal infections are not necessarily something every emergency physician encounters frequently in a busy environment focused appropriately on emergent care. But in primary care pediatrics, repeated exposure over years builds pattern recognition around these presentations.
That experience reinforced something important to me: The future value of AI in healthcare may not primarily be in writing notes after a visit. Its greatest value may be in helping patients and clinicians build a more complete clinical story before important clues are missed.
There is also another side to this conversation that few people discuss: AI creates an opportunity for medicine to become less siloed. Right now, medical knowledge is often trapped inside individual experiences. Over decades, doctors accumulate unique clinical insight that often disappears when they retire.
But what if medicine could learn collectively? What if anonymous patterns, atypical presentations, and nuanced patient experiences could help inform future questioning for everyone?
Patients everywhere deserve a fair opportunity to be asked thoughtful questions. Healthcare should not depend entirely on geography, network size, or whether someone happens to encounter the one physician who has seen a rare presentation before.
AI may finally allow medicine to scale thoughtful inquiry. Not perfect inquiry. Not infallible inquiry. But better inquiry. Used thoughtfully, AI may help physicians recover something modern healthcare increasingly struggles to protect: time for curiosity and deeper listening again.
And that matters.
Author’s Note: I use AI the same way I believe many physicians eventually will — not to replace human judgment, but to help organize information, explore ideas more thoroughly, and communicate more effectively. AI tools were used in the drafting and organization of this piece.
Editor Note: Dr. Russel McKenna is a pediatric physician and founder of Ask YADAYADA, an AI-assisted healthcare initiative focused on improving clinical questioning and thoughtful information gathering in non-emergent care. He has spent thousands of hours refining physician-guided AI workflows and has applied AI-assisted questioning approaches across thousands of patient encounters while advocating for AI as a tool that supports — rather than replaces — human clinical judgment.