Bob Abrahamson, Chief Marketing Officer
On April 28, JAMA Internal Medicine published an article comparing ChatGPT and physician answers to patient questions culled from Reddit’s AskDocs subreddit. A panel of 3 physicians judged responses on information quality and empathy. In both cases, the ChatGPT rated superior.
ChatGPT rating high on quality did not surprise me. Large Language Models (LLMs) can access so much more data, faster, than any human being. Via ongoing training, the AI is constantly learning and can present accurate, reliable information. Although my understanding is rudimentary, as evidenced by the explanation above, this makes sense. However, I was surprised by the empathy results. In fact, the team judged the ChatGPT results to be 9.8 time more empathetic than physician responses.
Many have explained this because of time pressures. ChatGPT has more time to craft an empathetic response than today’s busy physician. The average ChatGPT response was 211 words compared to the physician response average of 51 words. A driver of this study was an ongoing exploration of how AI can help address increasing workloads and burn out amongst providers. With an appreciation of the computing power and design of the LLM, it is not surprising that ChatGPT could generate more empathetic responses. But what do we do with this data point?
The Problem or the Solution
Digital Transformation, which inevitably will include leveraging all forms of AI, is about finding new ways to deliver value. It is not simply automating old processes. It is possible to read the results of this study and jump to the conclusion that we need to figure out how to use ChatGPT for drafting more empathetic physician answers to patient questions. Seems reasonable and it’s in the spirit of using the AI to assist the physician as opposed to replacing the physician. I wouldn’t disagree out of hand. But I believe we may not be looking at this the right way.
In my opinion, the core issue to address is the time crunch contributing to burn out for all types of providers. I believe doctors, nurses, therapists, etcetera are motivated in large part to help people as opposed to the profit motive. We need to figure out how to give them more space to show how they care, to be empathetic. The “quick” fix may be to leverage ChatGPT to draft a quality, empathetic response that a user can edit before sending to the patient. Wouldn’t that create more time for the doctor or nurse?
Perhaps, but if part of the quadruple aim is provider well-being, we must consider what provides satisfaction for them. I am reminded of my friend Paul, an ER nurse in chaotic city hospitals. He never stops moving during a shift. What makes him happy, besides the adrenalin rush, is teaching patients. He loves to draw on the dry erase board. Creating time by limiting the opportunities to be creative and connect with patients during those fleeting education moments would not make Paul happier on the job. In fact, it would desensitize him and hamper his ability to deliver excellent care.
Paul is not alone. As we continue to discover what AI is good at, we need to be sure to focus on how it can be used to improve the care experience for all stakeholders. Using AI to streamline and simplify Prior Authorization offloads a task most physicians prefer to avoid as much as possible. Robotic Process Automation enhancing claims processing is another use case that frees up physicians to focus on care. Yet, regarding provider-patient communication, the value of the study may lie in the fact that it is diagnosing an issue – how the current system impedes provider empathy – than in offering a solution. Our challenge as digital solution providers is to help streamline low value repetitive tasks and create more opportunities for providers to provide the high-quality care in manner that will enhance job satisfaction.