athenahealth BrandVoice: How Healthcare Leaders Can Build Trust In Clinical AI
Written by Dr. Nele Jessel, Chief Medical Officer, athenahealthToday’s AI-augmented era of healthcare has been years in the making. Throughout my career at the intersection of clinical practice and health infor...
Written by Dr. Nele Jessel, Chief Medical Officer, athenahealth
Today’s AI-augmented era of healthcare has been years in the making. Throughout my career at the intersection of clinical practice and health informatics, I’ve seen technology evolve from supporting individual tasks to reshaping how care teams make decisions and deliver care. What comes next will take continued innovation. It will also take something more difficult to engineer — clinicians’ trust, earned by giving them choice in how AI is used and a voice in how it works.
getty
Beyond the Adoption Numbers
More than 80% of physicians report using AI in their professional work, more than double the rate reported in 20231. That headline number gets cited frequently, and it should — it signals a genuine shift in how medicine is being practiced.
But the same AMA research that produced that figure also found something worth noting: 40% of physicians have balanced attitudes that are equally excited and concerned about AI, citing patient privacy and the integrity of the patient-physician relationship as their top concerns.
That tension — broad adoption alongside persistent, principled caution — is exactly what athenahealth’s AI on the Frontlines of Care research reflects. Among the clinicians we surveyed, 65% believe AI does more good than harm in the delivery of care. And in the same study, clinicians draw a sharp and deliberate line between AI as assistant and AI as decision-maker.
Seventy percent say building rapport with patients must remain human. Sixty-nine percent say comforting and reassuring patients must remain human.2
And when it comes to clinical decision support — the area where AI is advancing most rapidly and generating the most conversation — clinicians are consistent: they want AI to surface information, flag patterns, and reduce cognitive load.
Let it pull the relevant history, surface the pattern, catch the detail buried on page nine of a record — and leave the interpretation, the empathy, and the final decision with them. Responsible AI is a second set of eyes; the physician still makes the call.
Outside healthcare, the reflex is to praise AI for volume — more output, produced faster. Bring that yardstick into an exam room and it falls apart. What clinicians need is technology that creates time and space – the ability to be present in the exam room, without distractions, focused on the patient in front of them. AI earns its place in medicine by handing clinicians back their attention.
AI as Assistant, Not Decision-Maker
What our research reveals — and what I’ve observed directly — is that the practices where AI adoption is going well share a common characteristic: clinicians feel like it was built with them, not deployed at them.
Ambient documentation is the clearest example. Adoption is high, satisfaction is high, and the reason physicians cite most often is that it gives them back the ability to be present with patients during the day and with their families at night, instead of charting on both. It removes something burdensome without inserting something intrusive. The clinician remains fully in control of the clinical encounter. The AI earns its place by augmenting the physician instead of replacing them. That is the model, and it transfers.
At athenahealth, we have many structured programs designed to involve our customers in our development process — because we believe that clinicians who shape how tools are built adopt them differently — not just more willingly, but more effectively. When a physician has told us what matters to them in a workflow and can see that feedback reflected in the product, something changes in how they engage with it. It is the difference between a capability that gets used and one that gets released but never adopted. Where AI touches clinical judgment, we design for optionality and oversight as a matter of principle — and we are transparent about where and how AI is being used in our platform. Not because we lack confidence in the technology, but because we understand that trust in clinical AI is built the same way trust in any clinical tool is built: incrementally, through demonstrated reliability, with the physician’s judgment always in the loop.
A Responsibility to Independent Practices
I want to name something the aggregate data can obscure.
The physicians most at risk from poorly introduced clinical AI are often those with the least capacity to push back on it. In large health systems with dedicated informatics teams and implementation infrastructure, there is organizational bandwidth to evaluate new tools critically, raise concerns, and course-correct. In independent and small group practices — where a physician may be seeing a full panel, managing operations, and navigating prior authorization burden simultaneously — that bandwidth simply doesn’t exist.
When AI arrives in those environments without adequate introduction, transparency, or optionality, it doesn’t just underperform. It erodes the confidence of the very clinicians who might otherwise become its strongest advocates.
This is where the vendor community has a genuine responsibility that goes beyond product quality. These practices deserve the same rigor of design, the same commitment to explainability, and the same respect for clinical judgment as any large system deployment.
What Trustworthy AI Requires
AI is changing how medicine is practiced. I believe this fully and without ambivalence. The question is not whether clinical AI will reshape decision-making — it will, and in ways that will genuinely benefit patients. The question is whether physicians are partners in that transformation or subjects of it.
The data from our research suggests clinicians are moving thoughtfully and steadily toward broader AI adoption — a progression we’ve described as moving from First Win to Proving Value toward a New Normal of AI-infused care. That progression is healthy. It should not be rushed.
What it requires from vendors and health system leaders is straightforward, if not always easy.
Involve physicians before go-live, not after. The feedback you receive when a tool is already deployed is categorically different from the feedback that shapes a better tool. Design for oversight to be easy — a natural, respected part of the workflow rather than a workaround or a friction point. Make it possible for a clinician to ask why the AI surfaced a recommendation and receive a clear, honest answer. And be candid about what the technology cannot yet do reliably.
Clinical AI earns its place the way any trusted colleague does: by showing up consistently, being transparent about uncertainty, and demonstrating over time that its partnership makes the physician better — not obsolete.
That standard is not an obstacle to progress. It is the progress.
Dr. Nele Jessel is Chief Medical Officer at athenahealth and a co-chair of the athenaInstitute. She practiced pediatrics before moving into health informatics, where she has spent her career at the intersection of clinical care and technology.
1- American Medical Association 2026 Survey, February 2026
2- AI on the Frontlines of Care, athenahealth, December 2025