Ambient clinical AI has moved pretty quickly from something healthcare organizations were experimenting with to something many clinicians are using as part of their everyday workflow. That changes the conversation.

A year or two ago, much of the discussion was about what ambient AI could do. Could it listen to a patient encounter? Could it generate documentation? Could it reduce the time clinicians spend staring at a screen?

Now we have a more complicated set of questions. Is it actually delivering the value health systems expected? How should that value be measured? Are clinicians using it in the ways organizations anticipated? And what happens when the technology changes the interaction between clinicians, patients and families?

Those are some of the questions I’m looking forward to exploring at the Healthcare Innovation San Diego Summit on Sept. 28 during “Scaling Ambient Clinical AI: A CIO Playbook.”

The session brings together Stephen Gutierrez, CIO of Northeast Valley Health Center; Dr. Vivek Reddy, chief health informatics officer at University of Utah Health; and Dr. Mario Bialostozky, chief medical information officer and associate chief quality officer at Rady Children’s Hospital San Diego, with Adam Cherrington, founder of Aligned Health Advisory, serving as moderator.

And after hearing Adam and Dr. Bialostozky talk through the session, I think the conversation is going to be less about selling the promise of ambient AI and much more about what happens when that promise meets the realities of healthcare.

That’s where it gets interesting.

Beyond the AI implementation

Dr. Bialostozky shared a particularly interesting story about how Rady Children’s approached its initial evaluation of ambient solutions.

Rather than simply choosing one vendor, the organization structured its pilot so that different groups could try different solutions and then switch, allowing clinicians to experience the tools side by side.

That kind of real-world comparison is exactly the type of experience I think healthcare leaders can learn from. But the conversation didn’t stop with choosing a technology. It quickly moved to the question that comes after every implementation: What value are we actually getting from this?

For Rady Children’s, clinician satisfaction was an important starting point. The organization was looking at ambient AI in part as a way to address some of the administrative burden contributing to burnout.

But measuring that value isn’t necessarily straightforward. Dr. Bialostozky explained that his organization hasn’t necessarily seen a dramatic reduction in documentation time. And there’s a reason for that. Clinicians tend to use the technology with more complex patients, while simpler encounters may already have relatively streamlined documentation workflows.

That’s an important distinction. If you’re measuring success only by minutes saved, you could miss some of the other ways the technology may be changing the clinical encounter.

What happens when the technology gives clinicians their attention back?

One of the points I found particularly interesting was Dr. Bialostozky’s observation that ambient AI can allow clinicians to engage more directly with families because they aren’t spending the same amount of time looking at a computer.

That raises a bigger question about how we define productivity in healthcare.

•    Is success simply seeing more patients?
•    Is it reducing documentation time?
•    Is it reducing the need for traditional scribes?
•    Or is part of the value in giving clinicians more opportunity to actually engage with the patient or family sitting in front of them?

Those aren't necessarily the same measurement. And they may not produce the same answer when a health system's leadership team starts asking whether an AI investment is delivering the expected return.

As Dr. Bialostozky put it during our conversation, his organization is still trying to determine some of those answers. I actually think that is an important story to tell.

It’s OK to still be figuring it out

There has been an enormous amount of pressure on healthcare organizations to “do something” with AI. Adam made a point during the conversation that I think captures where the industry has been: “AI in healthcare has been fire. Are we ready? Did we even aim? And now it’s ready, aim, aim again.”

That evolution matters. Healthcare leaders are no longer simply deciding whether AI is coming. In many cases, they are already deploying it and learning what happens when the technology meets real clinical workflows.

That means the questions are becoming more nuanced.

• What works?
• What doesn't?
• What should we measure?
• What does the organization need to learn before expanding?
• And what are we still getting wrong?

I think those are much more useful questions than simply asking whether an organization has an AI strategy.

And then there’s trust

The conversation also took an unexpected turn that I found fascinating: the patient and family perspective. Adam raised the issue of patients increasingly using AI to find health information, while Dr. Bialostozky talked about what he is seeing with families who arrive with AI-generated information and, sometimes, considerably more anxiety about what might be wrong.

At the same time, clinicians are using AI tools to reduce administrative burdens and create more time for those same patients and families. There is an interesting tension there. AI may help clinicians spend more time looking at and talking to patients, while patients themselves are increasingly turning to AI for information before they ever walk through the door.

That creates a new question around trust. How do we make sure AI helps strengthen the relationship between patients and clinicians rather than complicating it? For me, that may be one of the most interesting parts of this conversation.

The human still matters

One thing came through clearly in the discussion: none of this is as simple as putting an AI tool into a workflow and declaring success.

There are questions about accuracy, privacy, security, adoption, clinician engagement and how the technology fits into the broader digital strategy. There are also questions about how patients and families perceive it. And ultimately, there is still a person responsible for making sense of what the technology produces.

Dr. Bialostozky put it plainly: “None of these tools are perfect. That’s why you need to have a human in the loop.”

That feels like an important place for this conversation to land.

Ambient AI has moved well beyond the “look what this technology can do” phase. Health systems are now gaining enough experience to ask harder questions about where it creates value, how that value should be measured and what needs to happen before they scale it further. That is exactly the conversation I’m hoping we’ll have in San Diego.

I’m looking forward to hearing from this panel of experts about what they’ve learned from the real-world experience of deploying ambient AI, where they’re seeing impact and where they’re still working through the answers.

There is still time to register for the Healthcare Innovation San Diego Summit on Sept. 28, and I hope you’ll join us for “Scaling Ambient Clinical AI: A CIO Playbook.”

I have a feeling this one will leave us with at least as many good questions as answers.

Would you like to join the conversation? Click here to register for the San Diego Summit.

Melinda Taschetta-Millane is Market Content Director of Healthcare Editorial, and Head of Content for Healthcare Innovation.