Healthcare AI often arrives with great expectations only to disappoint end-users as well as execs. Why is that?

By and large, U.S. hospitals and health systems do not have an AI adoption problem. But many—if not most—have an execution problem. 

This is to say they struggle to turn a set of installed AI tools into a measurable operational value. 

The unfinished status of the project is often understandable. That’s because the journey from investment to ROI involves the complex undertaking of applying “the right integration, workflow fit and execution discipline.”

So suggests David Chou, a Texas-based technology consultant who specializes in guiding executive decisions in healthcare. 

Chou is also a regular contributor to Forbes. The outlet published his thoughts on the AI adoption-to-execution challenge April 21. 

Peppering the piece with stats from the latest Qventus report on AI management by healthcare CIOs, Chou outlines four distinct if overlapping challenges. 

1. Healthcare CIOs and health systems must leave an AI pilot mindset behind and focus on AI execution. 

Real gains require integration into existing workflows and processes, Chou explains. 

“That means focusing on where AI fits within patient access, patient flow, revenue cycle, documentation and care operations, which are areas where workflow friction is visible, labor pressure is real and value is measurable,” he writes. “Data shows leaders now prioritize operational performance use cases, with more than 70% rating automated care operations platforms as critical to their 2026 objectives.”
 

2. AI governance is a barrier. 

Chou has observed that health systems lack a repeatable process for approving use cases, assessing risk, assigning ownership, setting data standards and establishing evidence before production. 

“This gap is why many initiatives stall,” he states, adding that four in five respondents struggle to measure AI ROI while 39% lack a clear process for benchmarking performance. 

“These are leadership and operating model issues vs technology shortages,” he maintains. 
 

3. AI’s impact on financial performance is similar to—yet different from—that of many other healthcare technologies. 

Measurable returns from AI are achieved in repeatable, operational areas tied to throughput, labor efficiency, reimbursement and productivity, Chou reminds. He names as examples patient scheduling, care access, documentation, coding, denials, prior authorization and operational efficiency. 

“Healthcare leaders will measure ROI through revenue, cost savings and improvements in patient outcomes and staff productivity, similar to the metrics health system leaders use for any technology investment.”
 

4. Healthcare CIOs do not want point solutions. 

Many CIOs are already pushing broader application rationalization strategies to reduce complexity, eliminate redundant tools and control costs, Chou reports. “This does not mean they should limit themselves to AI offerings already inside the current portfolio,” he emphasizes. 

“CIOs still need to stay open-minded about where real value will come from,” Chou writes. “The market is still taking shape, and it will take time to see which platforms and partners truly emerge as long-term winners.”