The use of AI is becoming increasingly common throughout healthcare, with applications of the technology ranging from transcribing appointment notes to supporting data analytics and diagnostics.
But patients’ and physicians’ opinions on AI are still relatively mixed. A survey published in July by JAMA Network Open found that patients generally perceived physicians as less competent, trustworthy and empathetic if they openly used AI.
Physicians have also expressed concerns regarding the use of AI in payer-related administrative processes, including prior authorization. In a recently released survey from the American Medical Association, 61% of physicians said they feared that payers’ use of unregulated AI was creating more prior authorization denials, which can cause delays in treatment and harm patients.
Two physicians recently joined Becker’s to discuss what areas of AI use they believe could use more clarity and what questions they still have about the technology.
Editor’s note: Responses have been lightly edited for clarity and length.
Question: Amid so much hype surrounding the implementation of AI in healthcare, what questions do you still have about it?
Marc Shelton, MD. Associate Chief Medical Officer at the University of Missouri Health System (Columbia): I follow the AI situation closely, partly because my son is a product manager for a very large corporation, and partly because I have always embraced innovative technologies that may help us take better care of our patients. I believe that AI has already helped us with search tasks, and some with transcription tasks. For now, it takes me about as much time to edit AI-word products as it does for me to generate it myself, so the efficiency is coming but not there yet. In medicine, more so than in most business and entertainment applications, we must have a high bar for accuracy, particularly when we start using AI more in our interpretation and decision-making applications. These will improve over time, but for now there is still room for improvement. I find that some of the claims from AI-sales folks and talking heads on TV are often ahead of actual function in the real world.
Kerri Layman, MD. Emergency Medicine Physician (Alexandria, Va.): As learners become reliant on AI, they gradually lose their background knowledge. Thus, how will they be able to provide the checks and balances to ensure the AI answer or tool is accurate so as to not harm patients?
