AI moving from hype to habit in physician workflows

A closer look at how physicians are experiencing AI in real time

Artificial intelligence has been a buzzword in healthcare for years. But now, the conversation is shifting—from what AI could do to what it’s actually doing. Across clinical and operational workflows, AI is beginning to show up in everyday care.

To better understand the shift, we asked clinicians in our 2025 Physician Sentiment Survey how their organization’s use of AI has evolved over the past year. The responses from 1,001 physicians nationwide offer a clear picture: AI adoption is uneven, but momentum is real. And in some areas, it’s already reshaping how care gets delivered.

AI adoption is gaining ground, but not for everyone

One-third (33%) of physicians said their organization currently uses AI, while 44% said they weren’t currently using AI and 23% said they weren’t sure—a striking signal that many clinicians may be using AI-powered tools without realizing it.

Not surprisingly, adoption varied by practice size. Among large practices (20+ full-time physicians), nearly 49% of physicians said they use AI. By contrast, only 22% of small practices (fewer than 7 physicians) and 23% of medium-sized practices (7–19 physicians) reported the same.

We also saw a demographic divide. Younger physicians were more likely to report using AI: 40% of those under the age of 40 said their organization uses AI, compared to 35% of those aged 40–64, and just 22% of those aged 65 and older.

We also saw some variation by geography. While roughly one-third of urban and suburban providers said their organizations use AI, about 28% of rural providers said the same. And across the board, many respondents simply weren’t sure about their organization’s use of AI tools. This suggests a gap in visibility and awareness, particularly in smaller or more distributed organizations, but it may also point to how integrated AI tools are in a lot of HIT software, blending into the background.

To get a closer look, we focused the rest of the survey on physicians who confirmed their organizations are already using AI and asked how its use has evolved across three key areas: documentation, clinical care and administrative operations.

AI in healthcare is transforming clinical documentation—fast

If AI has a breakout success story in healthcare, it’s documentation. Over two thirds, (68%) of physicians said their use of AI for clinical documentation such as ambient listening, AI scribe or smart transcription tools increased, while just 2% said it decreased.

This trend was widespread. Medium-sized practices saw the most growth, with 86% of physicians reporting increased use for clinical documentation. This could be because medium practices may have more resources to put toward artificial intelligence than smaller practices yet are more agile than larger organizations.

While sample sizes differ, early signs from rural practices suggest strong interest in documentation support tools. Among rural physicians already using AI, nearly 9 in 10 (89%) said their use of AI documentation tools increased over the past year. Urban and suburban physicians using AI also reported strong growth, with 66% indicating increased usage.

The appeal is clear. Documentation is a universal pain point. AI that reduces charting time, improves accuracy and fits into existing workflows delivers immediate value—no major behavior changes required. That makes it one of the most natural entry points for AI in clinical practice.

Clinical AI is scaling—but mostly in larger organizations

AI is also gaining ground in clinical care—supporting diagnostics, treatment recommendations, and risk scoring. Among physicians using AI, 46% said their organization increased its use of clinical AI over the past year. Only 3% reported a decrease.

The most growth came from large practices, where half of physicians reported increased clinical AI use. Small practices were close behind at 43%, while medium practices trailed at 35%. Because clinical AI often requires deeper EHR integration, more rigorous governance, and provider training, it tends to scale more readily in larger organizations that can invest in those foundations.

Physicians practicing in rural communities and larger metro areas reported similar year-over-year growth (46% and 48%, respectively). While growth appears consistent by geography, overall adoption still varies—particularly in rural settings—highlighting ongoing challenges around access, infrastructure and implementation support.

AI adoption for administrative tasks is more fragmented and less visible

AI’s role in automating administrative work is growing, too—though it’s happening mostly behind the scenes. Among physicians already using AI, 36% said their organization increased its use for tasks like medical billing, scheduling or prior authorizations. Only 4% saw a decline.

Growth rates were similar across large practices (40%) and small practices (39%), but substantially lower in medium-sized practices (24%). In fact, nearly 29% of respondents from medium-sized practices said they weren’t sure whether their administrative AI use had changed, indicating a gap in visibility or communication.

That lack of awareness isn’t limited to medium-sized practices: 19% of AI-using physicians overall said they’re unsure if their organization uses AI for administrative tasks, and another 19% said they don’t use it in this context.

Physicians not knowing about the presence of AI in administrative workflows isn’t necessarily a bad thing. In many cases, the most effective administrative AI works quietly in the background processing claims, routing tasks, or completing forms, without interrupting the clinician experience. That kind of invisibility means the technology is doing exactly what it’s supposed to: reducing the admin burden in healthcare so physicians can stay focused on patient care.

The opportunity now isn’t just to roll out more artificial intelligence. It’s to use it well.

Turning early momentum into sustained impact

AI is no longer a future concept. It’s already reshaping healthcare—just not evenly. Clinical documentation is the clearest win so far, with fast, widespread adoption. Clinical and administrative use is growing, but progress varies by organization size, setting and visibility.

The opportunity now isn’t just to roll out more artificial intelligence. It’s to use it well.

That means investing in tools that reduce friction—not add to it. It means building transparency and training, so physicians know what they’re using and why. And it means making sure practices of all sizes have a path forward, not just the biggest systems with the biggest budgets.

AI’s impact in healthcare won’t hinge on one big breakthrough. It’ll come from the accumulation of small, smart improvements, such as when a note writes itself, a task disappears, or a diagnosis gets just a little faster. The organizations focused on those everyday wins are the ones turning AI into a habit, not just a headline.