Many doctors worry that AI algorithms are pushing them to the sidelines. But others see these digital tools as partners in the quest to provide the best possible patient care.
By John Halamka, M.D., M.S., Diercks President, Mayo Clinic Platform and Paul Cerrato, MA, senior research analyst and communications specialist, Mayo Clinic Platform and a Professor at Northeastern University.
“The oldest and strongest emotion of mankind is fear, and the oldest and strongest kind of fear is fear of the unknown.” That observation, from science fiction writer H.C. Lovecraft also applies to many healthcare professionals who worry about being replaced or overshadowed by the latest AI models. Comments by prominent AI experts like Geoffrey Hinton have only fueled that worry: “I think if you work as a radiologist, you are like the coyote that's already over the edge of the cliff but hasn't yet looked down. People should stop training radiologists now. It's just completely obvious that within five years deep learning is going to do better than radiologists.” Although his prediction has yet to come true, it apparently has had an impact on medical students looking to choose a specialty.
One source states: “Medical students fear that AI will replace diagnostic radiologists to such an extent that they are choosing not to pursue careers in radiology.” That includes 20% of Canadian medical students and 50% of UK students. Similarly, 44% of American students say they would think twice about radiology. And it’s not just radiology that has many students and practicing physicians concerned about their future. It has spread to a wide variety of physician specialties, but their belief that AI could replace them ranges from 6-78% across 40 studies. And what’s especially revealing in those surveys indicate that the less physicians know about AI, the greater their fear about being replaced by an algorithm.
With such concerns in mind, MIT and the American Medical Association organized an event called “AI as a Catalyst”. Leo Celi, M.D., a senior research scientist at MIT and associate professor of medicine at Harvard Medical School, sent us a summary of their conclusions, many of which are listed below:
Key Takeaways
- The Physician's Role Must Evolve: AI's cognitive offloading requires physicians to "upload" new skills. The core question is whether this means becoming system architects and change agents, or if the profession will split into "physician scientists" and AI-assisted advanced practice providers (APPs).
- AI Can Be a Tool or a Threat: AI can free up time for patient care. However, biased tools can pose a significant risk if used as a replacement for critical thinking, especially by less-trained clinicians.
- Education is the Lever for Change: Medical education must be redesigned to foster critical thinking, systems design, and advocacy. This requires moving beyond incremental changes to radical curriculum shifts, like a first-year course that teaches critical evaluation of AI tools.
The Problem: AI's Impact on the Physician's Role
- AI's cognitive offloading forces a re-evaluation of the physician's role.
- Residents are concerned about what new skills to "upload" to remain relevant.
- Physicians are being asked to use tools whose real-world performance in diverse populations is simply unknown. FDA clearance for AI-based clinical tools lags far behind deployment, and post-market surveillance is weak. Organizations such as CHAI, the Health AI Partnership (HAIP), and Mayo Clinic Platform are actively engaged to ensure there is pre-market qualification of AI and post-market surveillance.
It’s unlikely we’ll see Dr. AI replacing physicians anytime in the near future, but we do need to stay abreast of developments in digital health and remain actively involved in the development and deployment of these tools.
Share this:
