From hype to practice: Urologists assess AI's real impact

When artificial intelligence (AI) first entered the clinical landscape, its potential painted a promising picture: one where clinicians could spend more time with their patients and less time on routine administrative tasks, helping to improve efficiency, reduce burnout, and bring the joy back to medicine. As these tools become more routine, the question remains: Has AI delivered on its expectations?

A recent study published in JAMA offered some insights, suggesting that AI scribe adoption had only a modest impact. The longitudinal cohort study, which included 8581 clinicians and 1809 AI scribe adopters, found that AI scribe adoption was associated with daily reductions of 13.4 minutes in electronic health record (EHR) use and 16 minutes in documentation time, as well as a modest increase of 0.49 patient visits per week.¹ They also found that EHR time outside of work hours did not change significantly. According to the authors, these findings suggest the need to better understand how these tools are actually impacting clinicians’ day-to-day workflow.²

Urology Times® sat down with several experts to get their thoughts on how AI is being integrated into their practice and whether those efforts have translated into meaningful benefits. Although these urologists agreed that AI has the potential to streamline administrative work, their experiences suggest the technology remains a work in progress—one that offers clear advantages but also introduces new challenges.

Jennifer U. Miles-Thomas, MD, MBA, of Northwestern Medicine, told us that although she felt optimistic about the promise of AI-based clinical tools when they were first introduced, her expectations have shifted over time.

“I had a very positive outlook because I knew there were changes that could be made to make the day-to-day a little bit easier, [for example,] to capture more information [and] not always rely on my own memory or timing of how to document,” she explained. “I had a very positive initial expectation. As time progressed, it became more realistic about what the software could do and the regulatory and liability issues associated with it.”

Adam B. Murphy, MD, MBA, MSCI, also of Northwestern Medicine, echoed this sentiment, explaining the impact of AI as a redistribution of the work, rather than a reduction. Because he cares for many patients with low health literacy, he noted that much of the “trust-building” conversation does not belong in the medical record, requiring additional editing of AI-generated notes.

“What is happening is I'm having to edit out a lot of the information that I think is extraneous to the medical record,” he said. “I'm also having to correct when it confuses my voice with my patient's voice, or when it confuses the third person—a caretaker, wife, husband, partner, or child—from the patients’. It is redistributing the work to a degree, but whether it's making me more efficient, I'm not sure yet.”

Where AI is helpful, and where it falls short

Although discussions around AI have provided utopian aspirations of its potential, Giovanni Cacciamani, MD, FEBU, urologist and director of the Artificial Intelligence Center for Surgical and Clinical Applications in Urology at the University of Southern California, says the expectations should remain grounded in reality. Rather than solving the issue of workforce shortages, Cacciamani sees AI as a tool that can alleviate some of the administrative burden attributed to increasing patient demand.

Related: Practical uses of AI across the urologic care continuum

“There is a lot of misunderstanding about the potential of artificial intelligence because it seems to be a cure for everything in medicine, which is not true,” he noted. “At the end of the day, we are living in a peculiar moment. The number of physicians is decreasing, and at the same time, the population is aging, thank God. But with aging, of course, we have more chronic diseases… [That’s] adding burden on a system that is already very stretched. AI can be a solution for trying to fix this situation, for decreasing the workload, and improving performance. It can be a perfect ally for us for decreasing the repetitive tasks that we do all the time.”

Miles-Thomas similarly emphasized that AI's greatest value lies in helping physicians reclaim time for patient care during the encounter itself.

“I think [the reason] why most of us go into medicine is to connect with people and solve a problem. They're coming to us with a problem that they've been unable to solve, or they can't solve, and through our education, we have some solutions. It doesn't mean it's the only solution, but we have some solutions that we're offering,” she explained. “I think one of the major limitations in medicine is time. There are always more patients than there are people who are able to spend the time or provide the resources of care. What AI is giving back is it's removing some of those administrative duties that don't need to be done by certain licenses.”

Related: The Expert APProach: Digital Health, AI, and the Evolving Role of APPs in Urology, with Jennifer Bepple, MD, MMCi

She specifically emphasized the value of ambient listening tools in allowing her to be more present during her patient encounters.

“The tools that I'm currently using include ambient scribes. I can tell you that it's made a world of difference. I do feel that I am able to spend more time with the patient. I can hear the story without interrupting, because my brain is now only focused on communicating with that patient and putting together what their clinical picture really looks like, vs trying to recall every facet of the conversation for documentation and billing and coding purposes,” she added.

Murphy also acknowledged that AI can provide meaningful benefits in certain settings, particularly when documentation demands compete with patient interaction.

“It's a plus in time-limited settings,” he argued. “It's a benefit for people who may not be so great with typing and seeing patients at the same time, so maybe older providers will like it better than younger providers.”

However, he also cautioned that current speech recognition technology has limitations in clinical scenarios involving multiple speakers.

“It may be hard for people who take care of a lot of patients with caretakers or maybe even people with racially concordant patients who have similar vocal patterns,” he added. “Learning how to distinguish the patient from the provider will be a big deal.”

Looking ahead, the experts agreed that AI's long-term success will depend on thoughtful implementation. Cacciamani emphasized that health systems should adopt AI strategically, prioritizing tools that demonstrably improve patient care rather than embracing innovation for its own sake.

“The main thing is the health care system has to open its eyes,” he said. “They have to be open to innovation but also be aware that not all the solutions are needed. [They should] focus on the AI solution that can ultimately provide better care.”

A pivotal moment in health care: What AI can and should do

This brings us to the current moment. AI tools are no longer an aspiration—they are being implemented at every level of the patient experience. Where we go from here will shape how health care is delivered, and it’s essential that integration keeps the human experience at the forefront.

“We're standing at this pivotal moment in medicine and health care,” said Miles-Thomas. “We're no longer just saying what we can do, but we're really trying to redefine the care itself, what we should do, who should do it, and we're trying to create a new health care that is organized, provides access, and delivers the appropriate care to patients.”

The choice now isn’t whether to use AI, but where to implement it and how. As the experts explained, these tools can provide immense benefit in certain situations, but there are limitations and potential pain points that must be considered.

Murphy emphasized the importance of this decision point: AI tools may never fully deliver on the aspiration to solve workforce shortages or eliminate burnout, but they shouldn’t become an added burden that further diminishes physicians’ desire to practice.

“We are at this great tipping point in society where artificial intelligence is going to help in lots of ways, but it's going to have pain points,” Murphy concluded. “I hope that providers are patient during that process. I hope that developers are very aware of burnout and that their technologies should not cause burnout. Marketing sometimes pushes things over physician and patient experience, and I hope that we don't risk [burnout] getting worse for the sake of integrating AI and saving the hospital system money vs patients getting better and providers being willing to stay in medicine.”

REFERENCES