I have never been shy about stating my belief that administrative burdens play a major role in the current tsunami of provider burnout. In fact, I would go so far as to say that administrative burdens are the #1 contributor to burnout.
I am heartened, however, by generative artificial intelligence (generative AI) tools that show promise for alleviating some of the load. We may soon see generative AI applications reliably lessen the burden created by documentation and other critical—but onerous—processes.
Still, we must be patient for a little while longer.
Earlier this year, I urged physicians to exercise cautious optimism about generative AI. It was not completely trustworthy then, and it is not completely trustworthy now. Nonetheless, as current limitations and problems with generative AI begin to be addressed, we will likely see more opportunities for its rapid adoption into practice workflows.
Actions to bring relief
Before talking about what generative AI might be able to do for physician workflows, we must first be clear about the difference between traditional AI and generative AI. In a nutshell:
- Traditional AI has existed for years and is great for pattern recognition. It is really good at following specific, preprogrammed rules to do specific tasks. Often, it is used to analyze data, such as looking at clinical patterns to determine which patients may be at rising risk for developing Type 2 diabetes. What it does not do is create anything that is brand-new.
- Generative AI is often considered the “next generation” of AI. In comparison to traditional AI, it can create brand-new content. For example, it could potentially be used to help providers write first drafts of email responses to patients’ questions.
If we consider AI tools holistically—whether traditional AI tools, generative AI tools, or a combination of the two—there are glimpses of an early positive impact on provider burnout. The caveat is that their benefits are only realized when providers have control over their selection and use, and providers actually leverage them. In other words, the use of AI tools cannot become just another mandated burden!
In addition, we must be wary not to allow other processes or work demands to “steal back” the time that could be liberated through generative AI adoption. Filling the time saved with other chores negates the benefit.
As I see it, two things are necessary before generative AI tools will truly offer administrative relief to providers on a larger scale:
- Maturity. Generative AI tools need time to mature—much like medical students growing into attending physicians. Fortunately, this is already happening. Ongoing work to address generative AI “hallucinations” and other shortfalls will undoubtedly help further develop the technology.
- Physician leadership. Physicians need to be integral to the adoption of generative AI solutions within their organizations. This may be easier for independent physicians who have more control over their own practices and workflows. Thus, we may see a sharp rise in adoption first within independent practices. Although generative AI adoption may be slower among employed physicians, it will happen. The key to successful adoption is for employed physicians to find their voice and press for action within their health systems. They should help inform decisions about investing in those generative AI tools that can best reduce their administrative burdens.
If these two things happen, there is a real future in leveraging generative AI to reduce provider burnout. These tools could help liberate physicians, nurses, and other providers from many of the burdensome administrative processes that have been thrust upon them for decades.
I am excited to see what the future holds.