Trust is a major theme flowing through many of the conversations healthcare leaders are having about how to safely and effectively incorporate AI into the field, pointed out Joel Gordon, chief medical information officer at UW Health in Wisconsin.
He made this observation during an interview this week at the Reuters Digital Health conference in Nashville.
Public trust is fragile — and one high-profile failure could stall progress for years, Gordon noted.
“Whether it’s gene therapy or whatever it might be, we have to historically reflect on things that we assumed we had trust for — but we didn’t build or gain the trust. And then something bad happened, and the floor fell out underneath it — for buy-in, for funding, for the ability to move the science along — and then we got blunted in growth. We can’t let that happen,” he declared.
It’s for this reason that he believes healthcare leaders have to prioritize AI governance and transparency.
Individual health systems and other organizations have established governance frameworks and clear rules of the road for AI use — but these efforts are still lacking at a national level, Gordon remarked.
When it comes to figuring out how to best govern healthcare AI, he said that the industry needs more collaborative learning instead of redundant research.
In his eyes, there needs to be more learning consortiums. He described these as collaborative groups involving various health systems, in which they work together to align research methods, goals and data frameworks to accelerate AI progress and reduce duplicate efforts.
“There is a bit of an opportunity for us to think about learning collegiums that have the same way of looking at data and the same ideas of where we’re trying to go collectively as an industry. We’re in the infancy, and I think it’s important that we recognize that. If we think we’ve had kind of a quick last 18 months — the next two or three years are gonna be amazingly, blisteringly fast with what we’re going to learn,” Gordon stated.
As healthcare providers continue to navigate this process, it’s important to remember that metrics and utilization matter more than flashy headlines.
Gordon noted that he often sees headlines that celebrate the speed and scale of AI rollouts, such as highlighting that 25,000 doctors went live with a tool in 10 months. But to him, this misses the point.
“That’s cool, but we’re not looking at the quality of the outcomes on the sides of all those different perspectives — billing, safety, patient education, continuity of record, routing of the documentation — and all of those things really do matter in the end,” he remarked.
Many hospitals tout their successful AI deployments — but real usage data, such as frequency and distribution across users, is often missing, Gordon added.
Overall, he thinks the industry should prioritize trust, collaboration and real-world outcomes to ensure AI delivers lasting value in healthcare.
Photo: Dmitrii_Guzhanin, Getty Images
