LAS VEGAS – If there's one thing everyone can agree on about artificial intelligence in healthcare, it's that everyone is still unsure about how to manage its many challenges and opportunities, and just where it's all headed.
It's a huge and complex and sometimes mysterious topic that's still new to a lot of people. It's transformative, and it's moving faster and evolving by the day.
"We're all gonna feel a little stupid sometimes if we wanna get this right," said HIMSS25 AI in Healthcare Forum co-emcee Brian Spisak on Monday morning.
Health system leaders are seeking insights and clarity as they chart strategies to use AI to optimize their financial and operational processes, create value, reduce documentation burden and streamline workflows – and ultimately improve patient outcomes and enable value-based care.
They're also grappling with the challenges: How to ensure transparency, explainability and equity of models and their outputs. How to protect patient safety as AI is applied in clinical settings. How to manage very real workforce ramifications.
In his opening keynote, "AI Policy Considerations and Practical Adoption Strategies for Transformative Technologies," Dennis Chornenky, chief AI advisor at UC Davis Health, explored those topics and more.
Chornenky – read our recent two-partprofile of him – pointed to a future where artificial intelligence has changed more about the day-to-day work of healthcare delivery than many might be expecting, even in these heady days of hope and hype.
As humans, "our minds are wired to be able to predict incremental change pretty well," he said. "What we're not very good at predicting is change that tends to jump forward a little bit more quickly: exponential change, or change by orders of magnitude.
"The environment we're in right now is one where AI's capabilities and potential impact really are advancing at an exponential rate," he said. "Over the next several years, we can expect, probably each of the years ahead, for AI's cognitive reasoning capabilities to jump by not 10 times, but more like 100 times with each iterative cycle."
We've spent most of the past two years talking about, implementing and exploring the possibilities of large language models and generative AI. Now, in 2025, agentic AI is the next frontier for healthcare.
"This is an iteration forward that leverages large language models, but also adds a lot of other methodologies to address workflows," Chornenky said.
The most succinct description of the difference between the two? GenAI can create content, but agentic AI can take action – even learning and setting goals.
"You can almost think of it like what we used to talk about as robotic process automation, or RPA, combined with AI – and what the various opportunities are," said Chornenky.
The opportunities for increased efficiencies are enormous, he said. But so are the potential risks to the sorts of jobs many people in healthcare have gotten pretty used to performing for themselves in recent years.
As "intelligent automation" transforms work processes, with AI becoming "multimodal, both in terms of ingestion of data, but also output," many use cases that once depended on many steps to accomplish, or the input and expertise of several different people, will be accomplished with a single prompt.
"That begs the question of what happens to people in junior roles and [their] intellectual work?" asked Chornenky. "PowerPoint decks, research reports, and things like that. If all of that can be automated, how do future generations, our children, get access to more junior-level jobs in order to get experience to get to senior-level jobs?"
His prediction: "I think our kids are going to have to get very good at managing AI. They're going to have to be able to say, 'Hey, I can effectively manage 10 AI agents.' That sounds like a stronger value proposition to an employer, I think, than just saying, 'Hey, I'm just a human, and I'm smart.
"That is the reality that we're going to," he added. "It's a little scary, but something we have to think about. So teach your kids how to use ChatGPT effectively now."
And agentic AI is only the next step. After that, as many predict, will come AGI, or artificial general intelligence.
"If an agentic AI can start to fully replace jobs, we can think of AGI as a higher-order intelligence that can perhaps even start to replace the management of entire enterprises," said Chornenky.
"Imagine a higher-order intelligence that can manage 100 AI agents that really doesn't require a lot of human supervision. Maybe there's a human board of directors, maybe there's some human technicians helping to run all the systems, but the majority of the productivity of that enterprise is being done by AI. Do we continue to regulate those enterprises in the same ways that we have in the past?"
And that's not even touching the cybersecurity implications.
"We'll be using AI to test for vulnerabilities on our own systems. But you can imagine an arms race where AIs are going at it?"
Healthcare organizations need to create a new governance and adoption infrastructure for autonomous AI, "because these capabilities are just getting more and more powerful," said Chornenky. "Without the right adoption infrastructure, regulated sectors are going to find it very difficult to adopt these more powerful technologies, whether it's healthcare or government."
That "AI governance gap," with innovation happening too rapidly for regulations and regulators to keep up, could prevent healthcare organizations from adopting AI more quickly.
That creates a dilemma for CIOs too, "because they're under pressure to adopt the new tech from their boards, from their partners and from patients, I think, to a growing extent. But they've got to ensure safety. And how do they do that if they don't have the proper governance mechanisms set up?"
Those are all questions that will have to be answered in the future. And that future may arrive more quickly than many expect.
In the meantime, it's essential for healthcare workers to get more comfortable and familiar with AI. Because there will be impacts on the job market as the benefits of automating certain tasks and roles become harder to look past.
Echoing a common refrain of the AI age, Chornenky said it is not, in the near team, "going to replace managers. But it is going to replace managers who don't use AI."
Mike Miliard is executive editor of Healthcare IT News
Email the writer: mike.miliard@himssmedia.com
Healthcare IT News is a HIMSS publication.