AI will not transform healthcare on its own. The technology needs clinicians involved from the start, workflows that reduce burden, and a deliberate effort to protect the human connection in patient care, according to Dr. Chris DeRienzo, chief physician executive of the American Hospital Association and president of the AHA Health Research and Educational Trust.
"AI is just a technology - it doesn't need to achieve anything," DeRienzo said. "Like electricity, the steam engine and the internet before it, AI is clearly positioned to become the defining general purpose technology of our era. But no matter how good a technology gets, its achievements alone cannot fully define its ability to transform any aspect of human society."
DeRienzo, who has clinical experience at health systems nationwide, is scheduled to speak on two panels at the HIMSS AI in Healthcare Forum in San Diego next month, covering AI governance and benchmarking.
Speed, simplicity and trust drive adoption
When DeRienzo examines hospitals across the country, three qualities stand out as adoption drivers: speed of access, simplicity of use, and trust in the output. He pointed to ambient documentation as the clearest example of near-term value for healthcare AI.
"One reason I think ambulatory ambient scribing has exploded in use is because it removes substantial burden I saw get added to documentation through the paper-to-digital record conversion," he said. "I also don't know any doctors whose favorite part of practicing medicine is writing notes. So, a simple, fast, AI-enabled technology with output a clinician can trust that removes burden of manual documentation checks a lot of boxes."
DeRienzo said he has been surprised at how quickly people began using straightforward AI tools for health decisions, even when other variables remain unproven.
Regulatory systems and the human obstacle
Changing clinical practice is difficult inside regulatory and reimbursement frameworks that were not designed for rapid technology shifts. DeRienzo pointed to the substantial regulatory changes required in the early 2000s to drive digital health record adoption - a transformation that still has not reached full completion.
Beyond structural barriers, healthcare presents a deeper challenge. Technology has too often weakened the connection between patients and caregivers rather than strengthening it.
"Even if both the regulatory and reimbursement mechanisms were already in place to support broad AI-enabled transformation, healthcare is and will always be a uniquely human experience," DeRienzo said. "For at least a generation, we've too often allowed technology to disintermediate the human connection most central to giving and receiving care rather than enable it."
He called this "the single greatest challenge" for both healthcare professionals and health technology developers. The task is figuring out "how to thread the needle of technology with the fiber of our humanity."
The implementation rule: put clinicians at the table
DeRienzo offered direct guidance for healthcare leaders planning AI projects. Clinicians must be involved from the beginning - before the problem is even fully defined.
"If you are planning to implement a new, AI-enabled technology in a healthcare professional's workflow, do not start the project without at least one if not several of those professionals at the table," he said. "Even better is bringing them to the table before you think you've clarified the problem you're trying to solve, because they understand not only the problem but the tradeoffs impacting why it is the way it is today better than anyone who hasn't recently walked in their shoes."
He added a budgeting rule: if you are not putting at least 75% of time and resources into people and process - rather than the technology itself - the odds of sustained transformation drop sharply. For operations leaders evaluating AI for Healthcare Courses, that ratio underscores where the real implementation work lives.
Why this matters for healthcare operations and management
DeRienzo's 75% rule is a concrete budgeting principle, not a metaphor. For operations leaders, it means project plans that tilt heavily toward technology acquisition will likely fail. The work is in workflow redesign, clinician buy-in, and process change. Start there. The strongest near-term AI use cases - ambient scribing, administrative automation - succeed because they remove known friction. Identify the friction first, then match the tool to it, not the other way around.
Your membership also unlocks:
