As the healthcare industry stands at the precipice of a technological revolution, few figures are as well-positioned to analyze the shift as Dr. Robert Wachter. The Professor and Chair of the Department of Medicine at the University of California, San Francisco (UCSF), and author of the seminal work The Digital Doctor, has recently released a new book, A Giant Leap: How AI is Transforming Healthcare and What That Means for Our Future.
In a recent episode of KFF’s Business of Health, hosted by Charles N. Kahn III, Dr. Wachter explored the profound implications of Artificial Intelligence for patients and clinicians alike. The conversation moved beyond the typical hype cycle, dissecting how these tools are already changing the reality of medical practice, the risks inherent in their deployment, and the evolving role of the physician in an age of automated intelligence.
The Current Landscape: AI at the Point of Care
The transition from paper records to Electronic Health Records (EHR) a decade ago was fraught with challenges—workflow bottlenecks, physician burnout, and a disconnect between clinical data and patient interaction. According to Dr. Wachter, AI is not merely a second wave of digitization; it is a fundamental shift in how medicine is practiced.
At UCSF, the integration of AI is already visible in the daily routine of hospitalists. Dr. Wachter describes using AI-powered scribes to document patient encounters, allowing him to maintain eye contact rather than focusing on a screen. He utilizes AI tools to summarize massive, complex medical histories—records that he notes can sometimes be "longer than Moby Dick"—and even uses AI as a "curbside consult" to brainstorm diagnostic possibilities for complex cases.
"These are relatively typical uses," Dr. Wachter noted. "It is remarkable for a field that tends to be pretty sluggish in adopting these kinds of technologies."
Chronology of a Transformation: From EHR to Generative AI
To understand the current "AI moment," one must look back at the trajectory of health technology.
- The Early Digital Era (2010s): The federal government’s "High Tech Act" incentivized the rapid implementation of EHRs. While this digitized patient information, it often placed a "productivity paradox" on physicians, forcing them to spend more time on data entry and less on patient care.
- The "Digital Doctor" Phase: Dr. Wachter’s 2015 book documented the hope, hype, and eventual harm caused by poorly integrated digital systems. Patients gained access to portals but lacked the tools to interpret their own data, leading to a flood of confusing messages for doctors.
- The Generative AI Pivot (2022–Present): The release of ChatGPT on November 30, 2022, served as a "lightbulb moment" for the medical community. Unlike the static EHR systems of the past, generative AI offered the ability to synthesize unstructured data, provide subspecialty-level insights, and draft complex clinical notes in seconds.
Supporting Data and Clinical Reality
While the potential for AI is vast, Dr. Wachter emphasizes the necessity of "singles versus home runs." The industry is currently seeing success in "singles"—low-risk but high-utility applications like drafting notes, automating billing, and patient outreach.
In radiology and pathology, AI is already proving its worth. At UCSF, radiologists—the group arguably most vulnerable to job displacement—are actively requesting AI tools to help manage an "undoable" volume of scans. This data point challenges the narrative that AI will immediately replace clinicians. Instead, the evidence suggests that in the current climate of workforce shortages, AI serves as an essential collaborator.
However, the "home runs"—such as AI autonomously curing diseases or performing complex surgeries—remain firmly in the realm of hype. Drug development, while accelerated by AI in the discovery phase, still faces the same rigorous clinical trial and regulatory hurdles that have always governed the industry.
Official Perspectives on Risks and Governance
The most pressing concerns for medical leaders are not just technical, but structural. Dr. Wachter highlights several "less obvious" risks:
1. The Security and Misinformation Threat
"Deep fakes and security risks are probably the main things that keep me up at night," says Dr. Wachter. As AI becomes capable of mimicking trusted figures, the spread of medical disinformation could be catastrophic. Furthermore, the risk of hackers compromising life-critical systems—such as pacemakers or insulin delivery pumps—represents a new, high-stakes security frontier.
2. The Patient-Facing Gap
While clinicians are learning to "co-pilot" with AI, the public is often using generic chatbots for medical advice. These tools can "hallucinate" or provide generic advice that lacks the necessary context of a patient’s history. Dr. Wachter stresses that patient-facing AI tools must be built to be "doctorish"—asking follow-up questions and iterating until the patient has a clear, safe, and accurate path forward.
3. The Regulatory Goldilocks Problem
The current regulatory framework, anchored by the FDA, is designed for hardware and static software, not for fluid, evolving AI models. Dr. Wachter argues that the FDA is a "square peg in a round hole" for regulating AI decision-support tools. He calls for a more creative, multi-layered approach that involves health systems, professional boards, and perhaps even AI-monitoring-AI systems to ensure ongoing safety.
Implications for the Future: "Never Skilling" vs. Deskilling
A central question in the debate is how the next generation of physicians will be trained. Dr. Wachter advocates for a careful approach to medical education. He warns against "never skilling"—where students fail to learn the foundational diagnostic reasoning skills because they rely on AI from day one.
"If we stop training young physicians in physicianship and learning how to make a diagnosis and have judgment, essentially we’ll turn them back into laypeople," he warned. The current strategy at UCSF is to teach students to form their own hypotheses first, and then use AI as a second opinion to verify their work.
A "Golden Age" or a Fiscal Mirage?
Looking toward 2035 and 2040, Dr. Wachter envisions a physician who acts more like an "orchestra conductor"—a curator of information who manages complex, high-stakes decisions with the help of AI agents.
However, he remains clear-eyed about the financial incentives driving the adoption of these tools. If AI is used primarily to "create a better bill" for insurance companies or to maximize hospital profitability rather than improving clinical outcomes, the technology may fail to solve the systemic issues of cost and access that have plagued healthcare for decades.
"It would violate the rules of both politics and human nature to say powerful tools that can be used to deliver an ROI… won’t be used for that purpose," Dr. Wachter noted. The path to a "golden age" of medicine requires more than just better software; it requires a fundamental shift in how the industry is paid and incentivized.
As the discussion closed, the consensus was clear: AI is not a magic bullet, but it is a "giant leap" toward a more efficient and capable system. The ultimate success of this transition will depend on whether society can build a reservoir of trust, maintain the sanctity of human judgment in high-stakes moments, and ensure that the benefits of this technology are directed toward the patient in the bed, not just the bottom line.
About the Author
