By: Casey Ma, MBA/MPH ’26, Yale Ventures Associate, Panel Planning Committee

About

On Tuesday, February 24, Dr. Kavita Patel (Stanford University; Venture Partner, New Enterprise Associates; former Obama Administration policy leader), Dr. David Rosenthal (Yale School of Medicine; Venture Partner, AlleyCorp), and Dr. Ashwin Vasan (Yale University; Operating Partner, Commonweal Ventures; 44th New York City Health Commissioner) spoke on the topic of Health in an Intelligent Age, a virtual panel exploring how artificial intelligence and emerging technologies are reshaping healthcare.

Together, they offered a candid view of what is working today, what remains hype, and where governance, evidence, and equity must catch up to technological progress. 

Event Recap

AI adoption in healthcare is already real, but not primarily where the public narrative often places it. While popular discussions focus on AI “replacing clinicians,” the panel emphasized that the most widespread use cases today sit behind the clinician–patient relationship: revenue cycle management, coding, administrative automation, and health system operations. These areas are gaining momentum because they offer clear return on investment and integrate more readily into existing enterprise workflows.

The story isn’t that AI is replacing clinicians. It’s that it’s being adopted where the system already knows how to pay for it. The earliest wins are not clinical breakthroughs, but operational ones. That tells us less about what AI can do, and more about how healthcare actually works. — Panel insight

On the clinical front line, panelists highlighted two areas where adoption is moving quickly: ambient documentation (AI scribing) and clinical decision support. These tools are gaining traction because they meet clinicians where they work, reduce immediate burden, and improve documentation and access to evidence. By contrast, many “point solutions” struggle to scale when they require new workflows, extensive integration, or a high trust threshold.

Equity was discussed not as an abstract principle, but as an operational reality shaping outcomes. The panel described a growing gap between resource-rich academic centers and community-based or safety net settings—differences in infrastructure, tool access, and AI literacy that continue to widen as capital and innovation concentrate in select environments. This uneven adoption creates a dangerous dynamic: rapid advancement where adoption is easiest, while many of the settings where most care is delivered remain under-supported.

The panel also explored why adoption remains slow even when tools appear promising. Health systems face an overload of vendors and claims, while internal decision-making is shaped by cybersecurity risk, contracting cycles, and trust. A recurring point was the influence of incumbent platforms—especially large electronic health record vendors—in shaping the market. If a health system expects a capability to arrive through an established vendor roadmap, it may delay adopting new solutions, regardless of their innovation. For founders, this underscores an important reality: technical performance alone is not enough; distribution depends on trust, integration, and enterprise readiness.

Beyond implementation challenges, the discussion turned to evidence: how should clinical AI be evaluated in the real world? Panelists noted a growing argument that AI should sometimes be evaluated against everyday pragmatic care rather than ideal guideline-concordant care, given that many patients do not receive guideline-level treatment today. At the same time, they emphasized that claims of diagnostic superiority or autonomous clinical decision-making must be supported by rigorous proof, transparency, and ongoing monitoring—especially as tools move into areas involving complex judgment and patient context.

Another urgent tension discussed was the disintermediation of healthcare by consumer AI. While health systems remain cautious and regulated, patients and families are already turning to general-purpose AI tools for health information and mental health support. This creates a governance challenge that cannot be addressed by clinical institutions alone. Modern policy approaches will need to reflect how these tools are actually used, rather than relying on frameworks designed for traditional, visit-based care.

AI is moving faster outside the healthcare system than within it. Patients are already using these tools to make decisions, often without clinical oversight. The question is no longer whether this is happening—it’s whether our policies, institutions, and care models are prepared to respond. — Panel insight

Panelists discussed the need for updated regulatory and reimbursement frameworks that keep pace with technological change. They highlighted both progress and remaining gaps, including the need for clearer post-market surveillance expectations for AI systems and the risks of fragmented state-by-state regulation. At the same time, they pointed to a broader contradiction: efforts to promote innovation are occurring alongside pressures on the very public infrastructure—research capacity, public health systems, and safety net funding—that supports safe deployment. In their view, AI leadership must focus not only on enabling innovation, but also on building the evidentiary and governance foundations that allow adoption to scale responsibly and equitably.

The panel closed with a forward-looking reflection on medical education and the clinician–patient relationship. Speakers expressed cautious optimism that AI can be a liberating force—automating documentation and other low-value administrative work while returning time to bedside care and human connection. But that outcome is not guaranteed. Whether AI strengthens or erodes trust in healthcare will depend on leadership, incentives, and a shared vision that prioritizes safety, equity, and meaningful improvements in care delivery. 

Closing

Yale Ventures thanks Dr. Patel, Dr. Rosenthal, and Dr. Vasan for their thoughtful leadership, and thanks to our partners across Yale School of Public Health, Yale School of Medicine, AlleyCorp, and Commonwealth for supporting this dialogue.