Robert Wachter, MD, returned to the SHM Converge mainstage and took attendees on a tour of his perspective on artificial intelligence (AI). Drawing from the more than 100 interviews he conducted as part of the research for his book “A Giant Leap: How AI is Transforming Healthcare and What That Means for Our Future,” as well as from his experience as a leader in hospital medicine and as an advisor for several tech companies, he presented a case for optimism in AI, its transition from novelty to normalcy, and its implications for leaders.
He described the current state of hospital medicine to show the transformative role that AI can serve. Hospitalists work in a healthcare system that remains broken for many patients and clinicians. Some patients struggle to access sufficient care, and clinicians find themselves encumbered by administrative burden, including excessive documentation. Regrettably, the advent of the electronic health record did not fix these issues.
While making the case for optimism about AI, Dr. Wachter highlighted its potential and outlined its roles in chart summarization, clinical decision support, documentation, and billing. His colleagues have welcomed the rollout of AI tools, including scribes, chart summarization tools, and discharge summary drafts. He believes these tools save time and let clinicians spend more time communicating with patients. He also said he looks forward to seeing AI-powered clinical decision support integrated into the electronic health record and called widespread built-in guidance for clinical decisions a potential home run.
His conversations with leaders in AI included more than optimism, however. There were discussions about the real downsides of AI. Dr. Wachter categorized concerns about AI into two categories: diminishing concerns and growing concerns. He stated that as AI tool developers have continued to improve AI, he has observed fewer examples of concerns such as hallucinations, sycophancy, and bias. Growing concerns he mentioned included privacy, security, and disinformation. Unfortunately, some AI tools present another avenue in which Health Insurance Portability and Accountability Act (HIPAA) violations could occur, so he encouraged the audience to exercise caution when using clinical information.
He identified the spread of misinformation as his greatest concern. By presenting a deepfake video of himself on rounds, he gave the audience a tangible example of how convincing disinformation can be. In the video, an AI version of Dr. Wachter spoke to other clinicians in a hospital hallway and made statements that did not necessarily reflect his views.
In speaking directly to leaders, Dr. Wachter made the case for governance structures that are nimble in this rapidly changing landscape. He encouraged the creation of systems with adequate monitoring and oversight. He suggested identifying skill gaps and training staff. His recommendations to leaders had a common thread of speed over perfection. He drove this point home by sharing a quote from Dr. Gianrico Farrugia that stated, “The risk of not going fast enough is far greater than the risk of going too fast.”
He concluded with a lightning round on topics such as deskilling, medical education, and the much-discussed question of clinician job security in the era of AI. He offered a relatable example of deskilling by saying that if he lost his phone, he might never see his wife again because he relies on it to store her number. On medical education, he recommended curricula that teach learners how to serve effectively as the human in the loop. He urged learners to prioritize diagnostic reasoning and avoid deskilling from AI use.
In one of his final points, he showed a line representing the likelihood that AI could replace a clinician’s role. He said the specialties most at risk rely heavily on pattern recognition, such as radiology, pathology, and dermatology. He added that surgical specialties face the lowest risk of replacement by AI. Importantly, he placed hospital medicine between these two extremes, describing it as a complex specialty that requires clinical knowledge, emotional intelligence, and care coordination.
He also noted that AI will likely never achieve 100% accuracy, so humans will continue to play an essential role in clinical decision-making. He closed by reading an excerpt from the end of his recent book, which tells a story about communication between clinicians and a patient. In it, he described rounding with his team on a young patient who had limited verbal communication abilities. He reflected on the roles clinicians serve and the functions AI could support in that setting, ultimately concluding that patients “will need a doctor.”
Key Takeaways
- Given the choice, Dr. Wachter chooses optimism for the future of AI in healthcare.
- AI has, and will continue to be, less a novel technology and more a normal part of healthcare.
- Leaders should adopt AI tools while recognizing their limitations and potential.
Dr. James
Dr. James is an assistant professor of medicine at Emory University in Atlanta.