Mark Thomas, Chief Technology Officer, MRO.
In healthcare, AI tends to scale broken infrastructure, not fix it.
Investment in healthcare AI has never been higher, and the promises have never been bolder, yet a substantial share of initiatives stall the moment they leave the pilot phase. The reason is rarely the model or even the AI implementation. It’s what sits underneath it—the data, the workflows and the operational fabric the AI has to operate on. When that foundation is weak, AI amplifies every gap, inconsistency and blind spot. And it does so at light speed.
To get real, durable value from AI, healthcare leaders have to invert the usual sequence. The AI technology is not the starting point. We have to first look at the foundation.
Why Most AI Strategies Fall Short
Most AI strategies fail before they start because AI is often treated as the first step. Instead of building the conditions that allow AI to succeed, many organizations layer AI onto existing challenges that have gone unresolved. Below are just a few common barriers that undermine AI’s effectiveness:
• Low-Quality Data: AI is only as good as the data it uses—and in healthcare, when about 80% of clinical data is unstructured, data curation and validation are especially critical steps. When AI operates on poor or incomplete data, it generates flawed insights that can’t be trusted.
• Inconsistent Workflows: When processes vary, results do too. This makes AI outputs harder to standardize, validate and scale across the organization.
• Limited Interoperability: Siloed systems prevent AI from seeing the full picture, meaning it operates with gaps in context that weaken its effectiveness.
What AI Needs For Success
Real results come from building the operational foundation that allows AI to work: how clinical data is accessed, structured and delivered, along with the workflows that support it.
To get the infrastructure right, you need standardized workflows that can be scaled; experience with efficiently scaling large amounts of clinical data in a compliant manner; and interoperable systems that make data connected and accessible across the enterprise. Expertly managing clinical data at scale typically improves consistency, optimized workflows create more reliable processes and connected systems ensure AI has access to the quantities of data it needs. Together, these elements create the foundation AI needs to operate effectively.
Now that you have the right infrastructure in place, how do you put AI to use? Some organizations may see AI as a way to reduce the workforce by automating tasks traditionally handled by staff. While AI may take on certain routine or lower-complexity tasks, that doesn’t eliminate the need for people but changes how their experience is applied. AI is not a replacement for human expertise. In fact, it’s just the opposite: AI delivers the most value when it’s partnered with human judgment.
To leverage the expertise you have in an exponentially larger way, the key is to develop a process that enables humans and technology to collaborate. This approach—which we call “collaborative intelligence” at my company—is built on continuous feedback loops between human judgment and AI recommendations. These feedback loops enhance shared learning so that both humans and AI are consistently improving over time, leading to better outcomes that neither can achieve alone. With every iteration, organizations see outcomes that are more likely to be reliable and effective.
Turn Strategy Into Action
To make a measurable, lasting impact with AI, you need more than just an AI strategy. Leaders should focus on a few key priorities:
• Strengthen interoperability before scaling AI. To help your systems communicate more effectively, explore centralizing clinical data management with the goal of streamlining access to disparate data sources.
• Standardize workflows for consistency. Workflows should be optimized for efficiency. This could involve user-friendly software, centralized technology systems or standardized training that reinforces consistency.
• Make human expertise a critical part of the process. Instead of thinking about how you can replace workers’ tasks with AI, think about how your team’s knowledge can be paired with AI for better outcomes. When humans and AI work together, AI is primed for better accuracy while teams can adapt more effectively to disruptions and remove routine burdens.
• Measure success based on outcomes, not speed. While efficiency is a plus, outcomes are what matter most, especially in the high-stakes healthcare industry. If you focus on making improvements in accuracy, quality and decision-making, efficiency will naturally follow.
AI adoption in healthcare moves at the speed of trust, and trust is built on a foundation people can actually see and verify—not on the sophistication of the model on top.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
