ChatGPT In The Exam Room How Doctors Use AI
From Patient Care to Paperwork
It might sound like science fiction, but for some physicians, AI is becoming a regular part of patient consultations. Adam Rodman, an internal medicine physician and Harvard Medical School professor, openly uses ChatGPT when faced with a challenging case. He ensures patient privacy by not entering personal medical information, and in one instance, the interaction became a collaborative effort. "The patient herself was typing in and giving additional information to the chatbot," Rodman shared. "It was a three-way conversation with the two of us in ChatGPT."
Rodman is part of a growing number of healthcare professionals turning to AI. A recent survey from Elsevier Health reveals that 48% of doctors now use AI tools in their work, a significant jump from 26% the previous year. This trend is fueled by a booming market, with AI-enabled medical startups attracting billions in venture capital, as noted in Rock Health's data.
The AI Gold Rush and Its Pitfalls
The major players in AI are vying for a position in the medical field. OpenAI's Sam Altman has claimed ChatGPT is a superior diagnostician, while Microsoft's specialized model reportedly showed four times more accuracy than human doctors in solving case studies. This has led to an explosion of specialized medical AI startups, creating what some describe as an overwhelming marketplace. John Brownstein, Chief Innovation Officer at Boston Children's Hospital, says he is pitched by "dozens every week."
However, this rapid innovation comes with risks. The technology is still new and often lacks extensive field testing. During a demonstration, pathologist David Zhang showed how ChatGPT incorrectly identified a prostate biopsy image as ductal adenocarcinoma, a type of pancreatic cancer. The diagnosis was "far from accurate," he noted, highlighting the current limitations of general-purpose AI and the continued necessity of human expertise. "That's why we have a secure job," Zhang said. "We can make the foundation model do better."
Many hospitals are proceeding with caution, running extensive internal trials to vet these tools for both clinical efficacy and ethical concerns. Rebecca Mishuris of Mass General Brigham avoids "shiny object syndrome," stating many tools offer "solutions for problems I just don't have."
A Unanimous Verdict on Curing Burnout
While opinions on diagnostic AI vary, there is one application that has earned widespread praise: ambient listening devices. These AI-powered tools record conversations between doctors and patients, automatically generating summarized clinical notes. This technology is being hailed as a transformative solution to the pervasive problem of physician burnout.
Carl Dirks, a physician at St. Luke's in Kansas City, calls it a solution for "clinician burnout," noting that it limits the "cognitive drain" of constant note-taking. His colleague, Philip Payne, adds, "We're really trying to restore human-to-human connection. How do we get the computer out of the way so that the provider and the patient have a conversation rather than the provider sitting behind a keyboard and typing the whole time?"
Psychiatrist Farhan Hussain, who no longer has access to an ambient listening device, misses it dearly. "Otherwise, we really are taking notes the whole time," he said. "Like damn, I didn't go to med school to just become a scribe."
Patient acceptance appears high, with one July study showing nearly 75% of patients were comfortable with the technology, provided its use is disclosed.
The Great Divide: Enthusiasts vs. Skeptics
Despite the clear benefits in administrative areas, a significant divide is emerging within the medical community. While fields like cardiology and radiology are pioneering AI adoption, many physicians remain skeptical. According to Elsevier Health, 24% of clinicians don't use AI at all, even outside of work.
Concerns range from ethical implications to the risk of skill degradation. An August study found that doctors who regularly used AI for colonoscopies saw their unassisted detection rates for adenomas decline. Physician Jonathan Simon worries about the push for efficiency over care. While AI can speed up diagnoses and increase patient throughput, he cautions that the industry must prioritize responsible use.
"Mistakes might be rare," Simon warned, "but a rare mistake can destroy someone's life."
