1. AI adoption in healthcare has accelerated significantly over the past few years. From your perspective, what is the most critical skills gap emerging among healthcare professionals today?
The most critical gap is AI fluency. Adoption has significantly outpaced training. The American Medical Association found physician AI use jumped from 38% in 2023 to 81% in 2026, more than doubling in three years. But in the same survey, 88% of physicians said they're worried about losing clinical skills to AI, and that worry was sharpest among doctors with ten years or less in practice. So adoption is outpacing AI fluency, understanding how a tool was built, where it fails, what it was validated for, and how to supervise it rather than defer to it. Without that, you get adoption without judgment, which is the worst of both worlds.
2. Many clinicians are already using AI-powered tools in their daily workflows. What competencies should healthcare professionals develop to use these technologies effectively and responsibly?
The most important competency isn't technical knowledge of data science or programming. It's understanding the technology’s limitations and maintaining strong clinical judgment. AI cannot replace the latter; in fact, it should ideally be freeing up clinicians to spend more time with patients and focusing on complex cases.
In addition to fluency and clinical judgment, the competencies that matter are the human ones: critical thinking, workflow awareness, patient communication, and escalation judgment. Knowing when to trust a tool, when to question it, when to seek another opinion, and how to document the decision transparently.
There's a quieter risk underneath all this. Lean on AI too early or too passively, and the underlying skills risk fading or never fully forming. I've called this never-skilling and deskilling. The clinicians who stand to do well aren't the ones avoiding AI or surrendering to it. They're the ones who use AI as one more input, not the only input.
3. As AI becomes increasingly integrated into clinical decision-making, how can healthcare organisations ensure that clinicians maintain appropriate oversight rather than becoming overly reliant on automated recommendations?
An important principle is to design for oversight instead of bolting it on. This means including explainability and transparency, not just its output. Make it easy for clinicians to flag any performance issues and create feedback loops.
4. One of the growing concerns around AI is explainability. How important is it for healthcare professionals to understand the reasoning behind AI-generated outputs, and how can they communicate these insights to patients?
Explainability matters for clinicians, but not in the way, most people might assume. A clinician doesn't necessarily need to see the model's inner workings. They need enough to answer practical questions: What data trained this? What is it designed to detect? How confident is it here, and how do I sanity-check that against the patient in front of me?
For patients, the same principle applies. They need to know AI was involved, what it was used for, and that a human clinician is accountable for the decision. That last part is what actually builds trust.
5. Healthcare leaders are under pressure to deploy AI solutions while ensuring patient safety and regulatory compliance. What role does education play in balancing innovation with responsible governance?
Education is the bridge between ambition and safe implementation. Leaders can buy technology, but they cannot scale impact unless care teams trust and adopt the solutions.
The numbers show why this matters. By 2026, around 75% of US health systems had adopted at least one AI tool, but only 18% had both a documented AI strategy and a functioning governance group to manage it. So most organisations are running AI without the structures to govern it. That gap isn't a technology problem. It's a knowledge and process problem. Leaders need the fluency coupled with governance processes to accompany the technology.
6. In your work as Head of AI Advocacy at GE HealthCare, what misconceptions about AI do you encounter most frequently among healthcare professionals, and how can they be addressed?
One misconception is that AI is either a fix-all solution or a threat to professional identity. Neither is true, and clinicians themselves mostly know it. In Elsevier's Clinician of the Future 2026 survey, 80% of clinicians said they expect AI to become a critical assistant in decision-making within five to ten years, not a replacement for them. AI is a tool. When thoughtfully designed and integrated, it can improve productivity, support decisions, and reduce friction in workflows, but it still needs clinical context, human empathy, and oversight. The replace-or-resist framing is mostly noise from outside the clinic.
7. Beyond physicians, which healthcare roles or departments do you believe require the greatest investment in AI training and upskilling over the next five years?
Three groups get overlooked. First, nursing. Nurses deal with more AI-driven alerts and monitoring than almost anyone in the building, yet they're rarely first in line for training.
Second, the administrative and operational layer includes scheduling, coding, patient access, and the revenue cycle. That's where AI gets deployed quietly, and where a bad model touches thousands of patients before anyone clinical notices.
Third, and this is the one I'd prioritise most, health system leaders. The execution gap is stark right now. The industry is in the earliest stages of the shift from experimentation to operationalising the technology. To successfully complete this shift, leaders need to plan strategically for AI and deploy it efficiently and sustainably, and that’s why fluency is so important.
8. How should medical schools, nursing programs, and healthcare training institutions evolve their curricula to prepare future professionals for an AI-enabled healthcare environment?
AI fluency should become a core part of healthcare training, not an optional elective for a small group of enthusiasts. Students should learn the fundamentals of AI, data quality, bias, validation, privacy, safety, and patient communication alongside clinical reasoning and ethics.
Some countries are already treating this as core rather than optional. France made AI and digital health training mandatory for all health professionals from 2025-26, backed by 119 million euros to train half a million people. That's the level of seriousness this needs. You don't close a workforce gap with a one-off lecture and an elective.
9. Trust remains a major factor in AI adoption. What practical steps can healthcare organisations take to build confidence among clinicians, patients, and administrators when introducing AI-driven solutions?
Trust starts with transparency. Organisations should explain why a tool is being introduced, what problem it is intended to solve, how it will be monitored, and what role humans will continue to play. Clinicians should be involved early, not informed after decisions have already been made.
Organisations also need feedback loops. Users should be able to report concerns, see how performance is tracked, and understand how governance decisions are made. Patients, too, deserve clear explanations of when AI is being used in their care and how clinicians remain accountable for decisions.
10. As AI automates routine tasks, how do you see the relationship between technology and the human aspects of care evolving in the years ahead?
The best use of AI is to give clinicians more time to spend with patients, versus spending that time reviewing charts, taking notes, and diverting their attention between screens. That's the ideal. Picking up the repetitive, low-value work is where we are seeing a lot of impact and adoption: reducing repetitive clicks, summarising information, and taking notes in the background.
That is the opportunity: technology handling more of the routine burden while healthcare professionals focus on judgment, empathy, explanation, and trust. Those human qualities will become more important, not less. A patient doesn't remember how fast their scan was read. They remember whether their doctor looked them in the eye and listened. If AI buys back even a few of those minutes, that's the win that matters.
11. Through your work with GR4AI.Academy, you've emphasised the societal impact of AI. What lessons from broader AI education initiatives can healthcare organisations apply when developing workforce training programs?
GR4AI is a different kind of project for me. It's a non-profit book that teaches children aged 7 to 12 what AI actually is, built with an 8-year-old co-author, so the language stays honest. The connection to healthcare is more direct than it looks. The kids learning how AI works today are the patients who'll be asked to trust an AI-assisted diagnosis tomorrow, and some of them are the clinicians who'll be supervising these tools in fifteen to twenty years. Fluency built early is the foundation that the health system eventually inherits.
Teaching kids also forces the discipline to avoid jargon. The same goes for healthcare. Patients don’t want complex terminology—they want a clear explanation they can follow.
The second lesson is that access matters. With kids, you don't pre-select the "technical ones." Everyone gets to understand the thing shaping their world. The same should be true in a hospital. AI fluency should be accessible to everyone.
12. Looking globally, are there healthcare systems or organisations that you believe are setting a strong example in preparing their workforce for AI adoption? What can others learn from them?
The strongest examples treat AI adoption as a workforce transformation, not just a technology rollout. Singapore is the clearest case I'd point to. Its national HealthTech agency, Synapxe, isn't only deploying AI, it's investing in the people to run it. Synapxe says it is training and certifying more than 300 healthcare technology professionals in AI, and they set up a HEALIX Data and AI Academy to keep that pipeline going over the next three years.
For the AI fluency program we’ve developed, HelloAI, we’ve been intentional about bringing in participants from organisations including Mass General Brigham, Alliance Medical, the University of Medicine Essen, and others to discuss their practical advice and real-world AI transformation journeys.
Closing Question: If you could give one piece of advice to healthcare leaders embarking on their AI journey today, what would it be?
Start with people, not technology. Identify the clinical or operational problem you want to solve, involve the people closest to that problem, and invest in education before, during, and after deployment.
Reference:
• "Healthcare AI Hit 75% Adoption. Only 18% Is Governed," ALIGNMT AI, 18 April 2026. https://www.alignmt.ai/post/healthcare-ai-governance-gap-2026
• Elsevier, Clinician of the Future 2026. https://elsevier.io/cotf2026
• OECD, Scaling Artificial Intelligence in Health, 2026. https://doi.org/10.1787/a436e12d-en
• Synapxe, "AI Accelerate 2025," 16 June 2025. https://www.synapxe.sg/media-releases/artificial-intelligence/ai-accelerate-2025
