"Losing That Trust Can Be Devastating": Dr. Jen McKay on Building AI Governance Clinicians Actually Trust

Before every text, page, phone call, and nurse pulling a physician aside at the station gets filtered into a single decision, a human being has to sort it all out first. Dr. Jen McKay, CMIO at Google for Health, calls that clinician the human API. Most AI governance programs, she said, are built without that person in the room.

Where Governance Skips a Step

McKay has spent her career moving between the exam room and the technology that is supposed to support it. Before joining Google about four years ago, she was a CMIO at a large Midwest health system and practiced as an internal medicine hospitalist for more than two decades, including in rural South Dakota.

"AI governance right now has a tendency to skip over the need for clinical leadership," McKay said. She often gets pulled into projects that are struggling to get off the ground or to scale, and once a clinician joins the effort, something shifts. Organizations tend to bring physicians in carefully and respectfully, but often later in the process rather than at the start. Bringing a clinician in early, McKay said, "can really change how quickly you can go, and also how well you do it."

That timing matters most in moments that look small on paper but are not. A tool rolled out to a physician on call at two in the morning does not get evaluated on its merits. It gets evaluated on whether it costs time or attention at the worst possible moment. Get that wrong once, and the cost is not just one bad rollout. "Losing that trust out of the gates of your clinical community can be really devastating for doing more work," McKay said. Rebuilding it afterward can take far longer than earning it would have in the first place.

Give Attention Back

Asked directly whether AI will eventually replace doctors, McKay did not offer a defensive answer. Instead, she pointed to an older technology shift. Telephone operators disappeared, she said, but a full telecommunications industry grew in their place. She expects something similar in medicine.

For McKay, that shift already shows up in her own work. Research that once consumed most of a project's time now takes a fraction of it, a change she compared to what happened when card catalogs gave way to searchable libraries, freeing her to spend more time synthesizing ideas rather than hunting for sources.

The bigger shift, McKay said, is not about time at all. "I think if we get AI right, it is not just time we are going to give back to clinicians, it is their attention." She sees that attention, and the human connection it makes possible, as the actual core of good medicine, especially for a patient who is scared and does not yet understand what is happening to them.

Watch Where Clinicians Already Trust It

McKay's advice for a CMIO watching AI adoption outpace policy starts with attention, not restriction. When staff are already using AI tools on their own, without organizational endorsement, that is not simply a compliance gap to close. It is a signal pointing at a process that needs fixing. She compared it to the early days of smartphone cameras, when clinicians began photographing X-rays and sending them to colleagues long before any policy addressed it. The workaround revealed the real need.

That same visibility carries risk, which is why McKay pairs the advice with a second one: bring security and compliance into governance early, not after an incident forces the issue. Protected health information has a way of surfacing in places nobody planned for, and the partnership with those teams needs to exist before that happens.

Picking the Right Fights

Adopting AI because it is available is, in McKay's words, not the right way to approach it. The clearest wins come from high volume, high frequency, error-prone workflows, with clinical documentation as the standout example, which is part of why ambient listening tools have taken off as quickly as they have.

She also pointed CIOs toward a workforce shift that gets less attention than it deserves. Organizations seeing real traction with AI tend to be the ones that have grown and upskilled their IT teams to operate more like product managers than traditional support staff, capable of rapid iteration rather than periodic upgrade cycles.

None of it works, McKay's advice suggests, without the thing she named at the start. Governance built on trust and clinical leadership is what allows AI to give clinicians their attention back.

Dr. Jen McKay is CMIO at Google for Health.