Artificial intelligence moves quickly. Wisdom often takes longer to catch up. That may be why a conversation from the HIMSS 2026 mainstage remains on my mind months later.

Amid thousands of pilots, predictions and promises about AI, three health leaders brought something valuable to the conversation: ethics and experience. Hal Wolf, FHIMSS, Isaac Kohane, MD, PhD, and Ran Balicer, MD, MPH, PhD, have spent decades thinking about how information, technology and balanced human judgment can improve people’s health. Each approach challenges from a unique vantage point.

I had the privilege of moderating their HIMSS “Views from the Top” conversation about recognizing value when selecting AI applications. The subject may sound technical. The conversation was warm, personal and reflective. At its heart was a very human question: How do we make certain these extraordinary new capabilities actually improve people’s health?

Three Must-Hear Voices

The people leading this conversation matter. AI in health does not need more prediction for prediction’s sake. It needs experienced leaders who understand how innovation moves from possibility to policy, from data to decisions, and from scientific discovery to patients’ lives and into health professionals’ decision-making.

Hal Wolf, FHIMSS, sees health from a global systems perspective. As President and CEO of HIMSS, the world’s largest professional society for health informatics, he works with health systems, technology leaders, governments and policymakers around the world. Under his leadership, HIMSS has forged agreements with governments and public institutions to strengthen digital use and the infrastructure needed to make health information accessible and actionable.

Those relationships give Wolf a view few people have. He sees innovation across institutions, borders and policy environments. He also understands that technology does not transform health simply because it exists. Progress depends on whether policy, infrastructure, information and people can align to put innovation to work.

Isaac Kohane, MD, PhD, has spent his career asking difficult questions at the intersection of medicine, science, computation, ethics and patient care. As Chair of Biomedical Informatics at Harvard Medical School, the physician-scientist has helped shape the field that many health leaders are now trying to understand at extraordinary speed. At the foundation of his medical training and perspectives is his training in pediatric endocrinology.

His questions matter as much as the answers. What assumptions are hidden inside these systems? What happens when we use powerful technology to automate practices that were flawed to begin with? How do we distinguish an impressive computational result from something that creates genuine clinical value? Dr. Kohane keeps the focus on consequences: What happens when technological capability outpaces human understanding?

Ran Balicer, MD, MPD, PhD, sees patterns and possibilities in health information and understands how to put them to work. At Clalit Health Services, he works within one of the world’s largest integrated health organizations, where longitudinal information across millions of people can reveal connections that might never be apparent during an individual clinical encounter.

For Dr. Balicer, accumulating data is not just about knowing more. It is recognizing risk sooner and turning information into insight that leads to informed action. That can mean preventing a hospitalization, identifying someone who needs attention or giving a health professional information that could improve or save a life.

That is what made their HIMSS conversation so valuable. Rather than predicting an AI-defined future, they explored what experience is already teaching us: How to evaluate innovation thoughtfully, translate information into action and ensure that greater technological capability advances the health of the people and communities we serve. Here are 10 takeaway lessons from their conversation to carry forward and apply in planning.

HIMSS 2026 “Views From the Top” conversation with Hal Wolf, FHIMSS, Isaac Kohane, MD, PhD, and Ran Balicer, MD, PhD, moderated by Gil Bashe.

1. AI Is Already Here

Health organizations may still be debating their formal AI strategies while clinicians, employees and patients are already “vigilantes” using the technology. Dr. Kohane captured the tension well: Health systems can move too slowly and too quickly at the same time. Formal governance may proceed cautiously while people inside those same systems reach for readily available AI tools to solve problems in real time.

That changes the leadership question. It is no longer a question of whether AI will enter health.  That ship sailed long ago. The task is to understand how it is being used, where it adds value, where it creates risk and how leaders can guide its adoption responsibly.

2. Speed Still Needs Structure

Wolf has seen health technology move through transformational cycles before. When internet connectivity entered hospitals, clinicians wanted access before many institutions had built the infrastructure and security systems to support use at scale. People found workarounds because the need was pressing and governance had to catch up.

AI is moving much faster. The answer is not to freeze innovation until every question has been resolved. Health organizations need structures that allow people to move with greater confidence. Responsible governance should enable progress, not become another source of friction.

3. AI Requires Human Values

Dr. Balicer raised an idea that belongs in every room where an AI decision is being made: Algorithms contain opinions embedded in code. Every model reflects choices about what information matters, which outcomes are prioritized, what trade-offs are acceptable and how clinical success is defined.

We should worry about bias in data. We also need to look at the assumptions built into the technology itself. Selecting an AI system, therefore, requires more than asking whether it is accurate. Leaders need to understand whose priorities it reflects and whether those priorities align with the needs of health professionals, patients and public health systems.

4. Begin With the Problem

Anyone who walked the expansive HIMSS exhibit floor saw the investment companies are pouring into AI-supported technologies. The temptation is to begin with capability: Look what this can do. Wolf’s perspective points toward a better starting place: What health-system and patient-care problem are we trying to solve?

Health has accumulated decades of technology intended to create efficiency, sometimes producing another screen, another workflow, or another demand on a health professional’s time. AI cannot become the newest layer of complexity. Begin with the health problem, understand the outcome that matters and determine how success will be measured. Then decide whether AI can help.

5. Do Not Automate What Is Already Broken

Dr. Kohane offered one of the most memorable cautions of the conversation: AI could effectively pour concrete over some of the worst practices in medicine. That image stays with you because it captures the danger of assuming automation equals improvement.

A poorly designed workflow does not necessarily become better because it becomes faster. Misaligned incentives do not improve because intelligence is layered onto them. Fragmentation does not disappear because information moves more quickly. AI allows health to reconsider old processes rather than simply automate them.

6. Leaders Need to Get Their Hands Dirty

Dr. Kohane provided wonderfully practical counsel to executives: Use several AI models yourself. Ask them questions. Compare their responses. See where they agree and where they wander. Experience how remarkably confident an incorrect answer can sound.

Health leaders do not need to become computer scientists. They do need curiosity and enough direct experience to ask better questions. Leadership becomes difficult when our understanding of a transformative technology comes entirely through presentations, dashboards and other people’s interpretations.

7. Prediction Creates Responsibility

Dr. Balicer brought the conversation squarely into population health. AI increasingly allows health systems to identify people at risk before they arrive at a physician’s office. Imagine recognizing months earlier that someone may be headed toward hospitalization or a preventable complication. That is the promise health has discussed for decades: moving from reaction toward prevention.

The challenge arrives with the insight. What happens when a system identifies more people who could benefit from intervention than it has resources to help? Better prediction does not eliminate the challenges of access to care. It makes choices more visible. Knowing earlier also creates a responsibility to decide what we will do with that knowledge.

8. Give Time Back to Healers

Administrative AI matters. Reducing documentation burden, improving scheduling and making information easier to retrieve can return precious time to health professionals. The more important question is, how is that added time going to be used?  This is a critical decision that leaders and health professionals need to resolve.

If AI gives a physician another 30 minutes, does it create an opportunity to listen more closely to a patient, explain something more clearly or go home a little earlier after a difficult day?  Will institutions see this as an opportunity to schedule another two patient appointments? Does it reduce cognitive burden, or does the schedule simply absorb another appointment? The value of efficiency should ultimately be measured by what it gives back to people.

9. Move Beyond Pilots

The health system loves pilots. They let organizations explore, generate enthusiasm and manage risk. They can also become comfortable places to remain when the harder decision is determining what deserves to scale.

AI maturity requires choices. Which applications improve outcomes? Which reduce friction? What enables health professionals to make better decisions? Which secures wider trust? Which should be stopped? Pilots offer baseline experiential information. Progress comes when leaders are willing to act on what they learn and scale.

10. Technology Amplifies Our Choices

AI can find patterns that people might take months or years to see, summarize volumes of information that few clinicians have time to read and identify risk earlier than traditional systems allow. Those capabilities are extraordinary. They still cannot decide what kind of health system we want to create or when a diagnosis may overwhelm a patient who needs added time with a health professional.

That responsibility remains ours. Wolf’s experience reminds us that progress requires systems and policies capable of supporting innovation and the people within that institution. Dr. Kohane reminds us to question assumptions and subject technological capability to scientific and ethical scrutiny. Dr. Balicer shows what becomes possible when information is connected across populations and transformed into insight, enabling people to act sooner.

Their experience and perspectives do not provide every answer. They do something equally valuable by sharpening the questions health leaders must be asking. What problem are we solving? What evidence demonstrates value? What assumptions are embedded in the technology? Who benefits? Who might be overlooked? What will we do with the time, information and predictive capability AI returns to us?

These are leadership and governance questions more than technology questions. The conversation about AI in health is moving beyond whether machines can perform well at their tasks. We know they can (even with their false flags). Our opportunity is to connect that capability with human judgment and care, operational realities, and an understanding of the people we serve, whose lives depend on ethical and clinical decision-making.

Will AI Replace the Doctor?

That question is getting louder. Christina Farr recently brought Ezekiel (Zeke) Emanuel, MD, PhD, and John Whyte, MD, MPH, together on her podcast series, Lifers, to debate whether AI could replace physicians in significant areas of medicine by 2030.

Perhaps the more important question is not whether AI will replace health professionals. Wolfe and Drs. Kohane and Balicer offered a roadmap to explore that question: “It is what we should allow AI to replace, what we should insist it improve and what must remain fundamentally human.”

AI may eventually outperform people at an expanding number of clinical tasks. The measure of progress should be whether those capabilities improve health while giving health professionals greater capacity to do what people still need from people.

Technology will continue to become more informed. Our responsibility is to become wiser about how we use it, remembering that every data point ultimately represents a person counting on us to make the right choice.