Walk into any emergency room across the country and you’ll see familiar scenes: crowded waiting rooms, overstretched resources, and overworked physicians balancing patient care with substantial administrative burdens. Against this backdrop, physician burnout remains unacceptably high (48.2 percent of physicians), with clerical duties cited as the most common cause. Many physicians also spend significant time after working hours catching up on clinical documentation, commonly known as “pajama time.”
Recently, health care systems have turned to artificial intelligence (AI) to help reduce administrative burdens placed on clinicians. As a medical student at the University of Chicago, I’ve had the opportunity to observe this technology’s implementation first-hand. In October 2024, the UChicago Medicine health system completed a three-month pilot with 200 physicians and advanced practice providers using an AI powered clinical documentation platform called Abridge. The ambient listening tool, accessible via smartphone, records conversations with patients and generates a clinical note based on the content of the clinical encounter. Rather than splitting their cognitive bandwidth between documenting in the chart and listening to the conversation, clinicians can instead focus on the patient in front of them and provide patients with near-instant access to summaries and visit information.
The pilot showed promising results, with benefits for patients and providers alike. Patients reported increased satisfaction related to the concern shown by their provider, explanations given, and inclusion in their care. Among clinicians, those who reported providing undivided attention to patients rose from 49 percent to 90 percent and after-hours work declined. The platform also positively affected hospital workflows, with more completed appointments per clinical session. Based on these findings, the University of Chicago has since expanded ambient AI access to more than 600 providers, including emergency department and inpatient physicians. Although limited, research on other AI-based platforms suggests they can improve clinical documentation, increase physician productivity, and reduce pajama time.
In my clinical experiences, I have seen how Abridge has benefited both health care professionals and patients. It has improved the efficiency of visits while enabling more engaged and thoughtful interactions with patients. Observing attending physicians, I noticed that meeting with patients, reviewing and signing notes, and transitioning to the next visit have been seamless. Moreover, the use of this technology by attending physicians has not diminished my medical education at all. It has led to more opportunities for instruction and helped refine my understanding of documentation and care planning.
Some critics have raised concerns about patient privacy and confidentiality. While Abridge records health care conversations, these recordings are deleted as soon as the clinical note is generated and never kept any longer than seven days. Additionally, multiple notices inform patients of its use in clinics, and verbal consent is obtained prior to recording.
Others have noted the risk of inaccuracies or misinterpretations of conversations. These tools do make errors, some of which may be clinically meaningful. Additionally, AI tools may struggle with diverse speech patterns and accents, leading to further mistakes. As these technologies evolve, it remains imperative that physicians carefully review the accuracy of their notes as quality control. Furthermore, developers should curate diverse and multilingual data sources to train their platforms to promote their quality and accessibility.
With their involvement, physicians have an opportunity to help direct this paradigm shift in clinical medicine. While electronic health records brought many clear benefits, they also imposed significant administrative burdens and were implemented largely without clinician input. AI tools provide an opportunity to correct some of these foundational missteps if clinicians are engaged in guiding their responsible use. Without their involvement, the development of these technologies will still proceed, but in ways that could diminish their value and safety for patients and providers.
If implemented responsibly, AI has the potential to ease clinical burdens, enhance patient care, and even promote medical education. However, realizing that potential will require sustained collaboration among clinicians, developers, and health systems, grounded in a shared commitment to patient safety and care.
Michael Wakeman is a medical student.
