In health system boardrooms across the U.S., conversations about artificial intelligence have reached a fever pitch. Vendors are promising transformation and executives feel pressure to act, but initiatives are stalling with striking frequency.
It’s not that the models fail; the health systems around them weren’t ready. While some believe AI has been overhyped, CIOs and IT leaders tell a different story.
“The biggest myth in healthcare AI is the belief that it’s a magic pill that will fix everything,” said Luis Taveras, PhD, senior vice president and CIO of Philadelphia-based Jefferson Health. “AI often amplifies the challenges we may already face. One of the first pressure points is governance: AI requests pour in from every direction and without a strong, disciplined governance model, organizations can quickly lose control.”
In a quick reaction to this growing problem, some health systems have created governance models specifically for AI to manage and triage requests, but Dr. Taveras has taken another route. He suggests integrating AI into currently trusted frameworks and elevating the projects that work well. That way leadership teams can take the time they need to evaluate AI technologies and pilots while reducing risks threatening organizational integrity and patient safety.
“AI holds tremendous promise and will undoubtedly transform every corner of healthcare,” said Dr. Taveras. “But we need to approach it with intention; not by reacting to every vendor pitch, but by prioritizing solutions that address clearly defined needs and deliver measurable value. To truly optimize the benefits of AI, we must be methodical, thoughtful and willing to slow down long enough to make smart, sustainable decisions.”
Aimmon Lago, executive director of revenue cycle systems at Stanford (Calif.) Health Care, said the field has made a categorical error by treating AI as different from every technology that came before it. The fundamentals of IT governance are important, and ignoring them in favor of AI-specific frameworks is a form of hype capitulation.
“Many of the best practices of IT governance to create and sustain value still apply,” he said. “These basics include thorough understanding of the target problem, calculation of the present value of risk adjusted future benefits and sizing of the cost to implement and sustain.”
Large language models are powerful tools and can create value if they are evaluated objectively alongside other technologies to avoid cognitive bias and assure value. Brian Lancaster, senior vice president and CIO of Children’s Mercy in Kansas City, said governance failures often occur because of operational issues. When organizations skip the preliminary work of defining the problem they’re solving, they are setting the projects up to deliver demos but not results.
“The biggest myth in healthcare AI is that it is a solution in and of itself,” said Mr. Lancaster. “Like every major technology before it, it is an enabler rather than the answer. Value emerges when organizations first define the problem by identifying the clinical or operational pain point, the stakeholders affected and the outcomes that matter.”
Health systems need a disciplined process to validate data readiness, assess workflow fit and set measurable success criteria before testing a small pilot and then scaling enterprisewide.
“Without this problem-first approach, the technology becomes a tool in search of relevance and initiatives almost inevitably fail to deliver meaningful results,” said Mr. Lancaster.
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
