Hospitals will soon need “AI Physician Specialists.” Interesting perspective in npj Digital Medicine, stating that the next frontier in clinical AI isn’t building better AI models.

SOURCE

Will instead be a “new profession”: the Algorithmic Consultant. Workforce of providers trained to safely and effectively use and oversee AI. Jayson Marwaha, MD, MSc Gabriel Brat, MD, MPH, FACS

Tal vez, empezamos un website, con esta option:

🧠 Why this matters

Thousands of predictive and generative models are entering hospitals.

Yet clinicians are not trained to judge whether a model is reliable, biased, or safe for patient use.

Similar to how pharmacists manage medications, these specialists will:

• Help clinicians select and interpret AI models at the point of care
• Oversee institutional governance (validation, bias monitoring, model updates)
• Bridge data science, medicine, and ethics in daily clinical practice

🐝 The cost paradox

Good call out by the authors on practical tensions in clinical AI implementation:

• AI oversight isn’t free. Hiring domain experts to oversee and validate deployed models adds significant costs.

• Costs can offset savings. The more oversight required, the smaller the economic gain from AI.

• Yet removing oversight is a non-starter. Without clinicians-in-the-loop, errors can propagate system-wide.

• Result: success of AI in medicine depends on balancing efficiency gains with cost of safety.

• Key question: can health systems design governance structures that are scalable, without introducing the same cost pressures AI was meant to solve?

💡 Broader insight:

AI may change who does the cognitive work of care / how it’s performed (eg. from physicians alone to physicians + AI), but not how much cognitive work the system requires.

Because oversight itself becomes an added clinical function that requires cognitive effort, and must be staffed, trained, and budgeted like any other part of patient care.

📍 Stanford BMIR Colloquium this week

“Updating a Mathematical Model of Cervical Cancer Screening”

🗓️ Oct 30 | 12–1 PM PT | Tapao Hall (3180 Porter Dr, Palo Alto) + Zoom

🎙️ Dr. Michael Higgins, Stanford BMIR + Jonathan H. Chen

Revisiting one of the earliest examples of mathematical modeling in medicine (decades before “AI” existed) and what it reveals about the limits of today’s reasoning systems.

Shows how quantitative reasoning has always been central to clinical decision-making, even when the tools change.

📘 Early registration closes this week for Stanford’s AI in Healthcare Leadership and Strategy program (May 2026).

4-week hybrid, CME, Stanford certificate.

Will have faculty and guest faculty from Stanford, Harvard, FDA, Google, Microsoft, Roche, leading AI labs etc.

Including Adam Rodman, Emily Alsentzer, Eric Horvitz, Okan Ekinci, MD MBA, Jonathan H. Chen, Jacqueline Shreibati, MD, Vishnu Ravi, MD, Shantanu Nundy and more.