How to Turn AI Promise into Real Progress in Healthcare 

AI adoption in healthcare has accelerated, but the gap between enthusiasm and practical results is still wide. As health systems move into 2026, leaders are shifting their attention from theoretical potential to measurable impact. Recent insights from executives at HCA Healthcare, Beth Israel Lahey Health, AdventHealth, Duke Health, and CorroHealth reveal a more grounded view of what AI can realistically deliver today and what it will and will not solve in the years ahead. 

Where AI Is Already Making a Difference 

While healthcare faces competing technology priorities, several areas are showing reliable impact. Clinical and operational leaders agree that the most immediate gains come from reducing administrative burden for clinicians and staff. 

Tom Scoggins, MD, chief medical officer at AdventHealth DeLand, noted that electronic medical records promised efficiency but often created more work. He sees AI helping clinicians reclaim time, especially through technologies that simplify documentation or surface relevant clinical information. His team is piloting a system that reviews patient charts, generates problem lists, and supports decision making. It required adjustments early on, but clinicians soon identified unexpected ways the tool improved their workflow—evidence that frontline engagement reveals value that vendors and leaders may overlook. 

Jake O’Shea, MD, chief health information officer at HCA Healthcare, described similar gains. His team focuses on tools that keep “the human in the loop,” particularly for administrative tasks. HCA’s AI-enabled scheduling tool, for example, lightens the workload for nurse leaders. Instead of starting with a technology concept, O’Shea’s team starts with a frontline problem, determines whether AI can improve it, and then measures results through time saved and sustained clinical quality. 

Revenue cycle leaders see parallel opportunities. At Beth Israel Lahey Health, Vice President of Mid Revenue Cycle Keisha Downs is seeing early traction with autonomous and semi-autonomous coding. As documentation quality improves through ambient tools, coding teams can handle higher volumes with greater consistency. She points to measurable shifts in accounts receivable days and denial rates as early indicators of real progress. Productivity improves not because AI replaces coders, but because AI does the lowest-value work, allowing staff to focus on cases that require human expertise. 

Understanding What “Enterprise Ready” Actually Means 

One of the clearest insights from the panel was that AI adoption in healthcare is only successful when governance is strong and expectations are realistic. 

Rick Shannon, MD, chief quality officer and chief medical officer at Duke Health, stressed the importance of proof-of-concept work grounded in local data. Duke built its own curated data environment and in-house models over several years. Tools such as its deterioration index and sepsis prediction program are trusted precisely because they reflect Duke’s patient population and care patterns. Even then, the team carefully calibrates sensitivity to avoid alert fatigue and preserve clinical judgment. 

Scoggins reinforced the need for thoughtful oversight. AdventHealth created a formal AI governance structure led by a chief AI officer. This group reviews clinical validation, workflow fit, security, vendor readiness, and policy alignment. The committee moves quickly enough to support innovation but thoroughly enough to maintain safety. 

Across health systems, leaders echoed the same theme: enterprise-ready AI is not defined by technical sophistication alone. It must scale safely and integrate into real workflows. 

The Human Side of AI Adoption 

If technical readiness is one half of the equation, change management is the other. 

Downs described the anxiety many revenue cycle teams feel: that AI will replace rather than support them. She has seen promising technologies fail when they were rolled out too broadly or too quickly, eroding trust and requiring the entire effort to be restarted. Her most successful implementations focus on education, transparency, and small high-impact pilots. Staff need to understand that AI is there to eliminate repetitive work so they can deepen their expertise. 

Shannon added that clinicians’ skepticism is earned. Many entered medicine before the EHR era and remember promises that became burdens rather than solutions. New AI tools, no matter how impressive, are layered on top of systems that clinicians already find cumbersome. Leaders must acknowledge this history and build trust through consistent delivery of small, reliable wins. 

Jerilyn Morrissey, MD, chief medical officer at CorroHealth, emphasized that leaders also set the emotional tone. Healthcare is a high-pressure environment, and teams look to leaders for clarity and purpose. She sees patients themselves driving demand for AI-enabled improvements, especially tools that improve access and communication. Including patients in innovation efforts helps organizations stay grounded in what people actually need, rather than what technology can theoretically provide. 

Moving From Reactive to Proactive Care and Finance 

Beyond immediate efficiencies, AI adoption in healthcare is beginning to reshape long-term models of care and financial management. 

Shannon highlighted the reality many systems face after the Great Resignation: less experienced clinical workforces caring for increasingly complex patients. Duke Health’s initiatives illustrate how continuous surveillance, predictive tools, and rapid response teams can support earlier clinical intervention. The goal is not to automate clinical judgment but to strengthen it. 

On the financial side, Downs expects a significant shift from reactive denial management to proactive strategies. AI tools can flag missing documentation, predict denial likelihood, and detect payer patterns earlier than manual review ever could. This creates space for teams to intervene before revenue is at risk, rather than after the fact. 

Morrissey sees a similar change on the clinical side. Readmission risk prediction has existed for years, but AI enables more precise targeting and automated follow-up. When providers, payers, and patients share insight into early risk signals, new reimbursement models built around prevention become possible. 

The Mindset That Separates Leaders from Laggards 

At the close of the discussion, each panelist named one shift that will define successful AI adoption in healthcare. Their answers converged around a shared message. 

O’Shea argued that disciplined selection matters more than sheer volume. The organizations that win will not be the ones that deploy the most tools, but the ones that deploy the most trusted and scalable tools. 

Scoggins focused on people. Productivity gains should be reinvested in patient connection, not used as justification to add more tasks. AI is valuable only if it strengthens human relationships. 

Shannon offered a caution: AI cannot fix broken processes. Leaders must stabilize and standardize their workflows before applying advanced tools. Without that foundation, even the best technology will underperform. 

Morrissey summed it up with a call for constructive reimagination. AI requires leaders to rethink what current workflows should be and how AI can be aligned with the organization’s purpose. 

Downs emphasized partnership between people and technology. Success depends on teams that view AI as an ally, rather than a competitor. 

Taken together, these insights form a clear picture. AI adoption in healthcare is no longer about experimentation. It is about thoughtful, steady, well-governed progress. The leaders who treat AI as a practical capability—not a spectacle—are already seeing results.