A Digest of John Glaser’s Interview
Insights from John Glaser: Charting AI’s Practical Future in Healthcare
The latest episode of Practical AI in Healthcare featured John Glaser, a true innovator in healthcare IT, who discussed the evolving impact of AI on the industry. Across four decades of hands-on experience, from CIO at Partners Healthcare to executive roles at Siemens and Cerner, and now teaching at Harvard, Glaser brought a grounded perspective to what really moves the needle in healthcare AI.
The Arc of Technology Innovation
Glaser highlights how history repeats itself in waves of innovation. Every decade delivers a new paradigm: from mainframes and networked PCs to the web and mobile devices, each breakthrough sets the stage for new capabilities. Today, AI’s rise, primarily driven by large language models, is fueled by rapid advances in computing power, increased data availability, and the proliferation of practical business models. But Glaser cautions against expecting instant transformation: meaningful progress always takes time, and we are still in the early stages of AI’s real-world integration.
Beyond Hype: What AI Actually Delivers
A recurring theme in Glaser’s analysis is the difference between hype and reality. While AI’s buzz dominates headlines, actual productivity gains often occur in “unsexy” areas, such as prior authorization and utilization management—tasks that deliver real value but rarely garner attention. Glaser points out that when the risk of error is manageable, administrative use cases can provide fast ROI for health systems under financial stress. He predicts that such applications will scale quickly and that many dramatic changes are still to come as adoption becomes more widespread.
Domain-Specific vs. Generalist AI Models
One of Glaser’s key insights involves the importance of domain expertise. In healthcare, specialized AI models outperform generalist ones, especially where context and experience are most crucial. He draws a vivid analogy to clinical decision-making: just as a specialist sees through irrelevant data, AI models tuned for specific workflows are more trustworthy and efficient. This approach is crucial for administrative domains, but also increasingly for guiding clinical capacity and patient triage, supporting not only productivity but also care quality.
The Human Element: Literacy and Implementation
Transforming healthcare with AI isn’t just about technology—success depends on careful change management and staff “literacy.” Glaser emphasizes that the front lines (not just middle managers) need real-world exposure to AI tools, the authority to shape initiatives, and a say in evaluating results. Experience—not just education—drives adoption and helps professionals overcome fears about job loss or failure. Leadership must establish trust, communicate transparently, and focus on pilots that let teams see AI’s impact before scaling.
Building the Right Infrastructure and Governance
As with previous technology revolutions, the surrounding infrastructure is critical. Glaser notes that, just as cars needed roads and insurance, AI’s broad adoption relies on economic, legal, and technical frameworks. Currently, there is limited structured oversight when AI systems malfunction in healthcare—regulatory guardrails and voluntary industry standards must evolve to keep pace and prevent harm. The speed of technological development often outstrips society’s ability to adapt, emphasizing the importance of a balanced approach to governance.
Where to Look Next
For those seeking AI’s “killer apps” in healthcare, Glaser advises focusing on areas under financial and workflow stress. Administrative automation, more intelligent triage, and clinical capacity management are ripe for innovation. He predicts the success of AI will hinge less on “sexy” demos and more on solving persistent, system-wide problems. Finally, Glaser’s measured optimism shines through: “This is a remarkable time. This is remarkable technology… the way you deal with it is one step at a time.” Bite-sized experiments, practical pilots, and sober analysis will pave the way.