Abstract
Objectives
To qualitatively characterize barriers and facilitators to implementing and using an ambient scribe across a large academic medical center, as well as how ambient transcription reshapes clinicians’ perceptions of their work.
Materials and Methods
We conducted semistructured interviews with clinicians who participated in an ambient scribe pilot (n = 8) and the initial enterprise rollout (n = 16). We sought heterogeneity by specialty, note volume, burnout, and prior time-in-notes. Interviews (26-60 min) were recorded, transcribed, and analyzed thematically using a naturalistic, ethnographic approach informed by broad implementation considerations, and an analytic lens treating note sections as documentation “genres.”
Results
Clinicians described feeling more present with patients and greater satisfaction during visits. Fictions included overlong or underspecified sections (eg, History of Present Illness vs Assessment & Plan), unfamiliar formatting, and a perceived loss of “voice.” Participants discussed how they used documentation to personalize practice, demonstrate expertise, manage impressions with colleagues and supervisors, and communicate sensitive findings—activities not fully captured by efficiency metrics. Inpatient and procedure-heavy contexts reported limited benefit where documentation was already highly standardized.
Discussion
Early ambient scribe implementation produced recognizable benefits, but introduced new work to reconcile AI-drafted text with local documentation genres and audience-specific communication. Tailored prompts, onboarding, and peer support may reduce the need to revise artificial intelligence (AI)-generated text.
Conclusion
Ambient scribe adoption can enhance patient interactions and perceived efficiency while reshaping how clinicians express voice and expertise in notes. Implementation strategies attentive to documentation genre and audience may help align ambient scribe outputs with clinical communication needs.
Keywords: generative artificial intelligence, qualitative, evaluation, implementation, ambient scribe
Background and significance
Ambient scribes listen to the provider-patient conversation and use generative artificial intelligence (AI) to draft clinical documentation. In practice, providers enter the exam room, start the ambient scribe using either an app on their phone or a desktop computer, conduct their interview and exam, then stop the ambient scribe and leave the room. The ambient scribe documents a transcript of the conversation in addition to generating a clinical note (within 30 s of the end of the appointment). In our context, the clinician can generate and regenerate a note, changing the instructions they give to the ambient scribe about form and content. Alternatively, the note can be manually edited by the clinician. Once satisfied, the clinical note is entered into the electronic health record (EHR). Although the popular discourse on ambient scribes is mixed,1–3 the research strongly suggests that using ambient scribes reduces documentation burden4–21 and increases job satisfaction.4,9,11,13,19–24 However, measures of documentation burden and job satisfaction (ie, time in notes, clerical workload) may not account for important characteristics of documentation work.
Part of what clinicians do with their clinical documentation is “translation work,”25 or documenting (eg, making a clinical note) in such a way that it can be used by different audiences. Medical coders need to discern the International Classification of Diseases, Tenth Revision (ICD-10) codes from a clinical note; patients need an understandable and comprehensive After Visit Summary based on the clinical note. Ambient scribes have the ability to do this translation work for the clinician; in part, this may be why clinicians describe how one of the benefits of this technology is that it increases their “efficiency.”5,8,9,12,23,26,27 In prioritizing efficiency, clinicians may be pointing to what Tracey et al describe as the “logistification of documentation,” or how documentation can function to expedite the movement of people, resources, and/or information, through an infrastructure.25 The current discourse on the benefits of ambient scribes relies on conceptualizing the documentation work done by clinicians in part as translation work, and in part guided by logistification. However, we believe this discourse offers an incomplete understanding of documentation work.
A broader and more critical perspective is needed to understand additional ramifications of AI-mediated documentation. In addition to translation and logistification, documentation also serves to reinforce and reproduce the social world of clinicians (eg, their professional identity, interprofessional relationships, and clinician to clinician communication). These practices and conventions exist between and among clinicians in the real world, and can be read and recognized by clinicians in clinical documentation28–30; however, few studies have explored how ambient scribes impact these qualitative, sociotechnical, aspects of clinicians’ voices in the clinical note. Our qualitative findings provide some insight into clinicians’ perspectives on how they listen to the note, and think about their notes, when the note may no longer be in their voice.
Materials and methods
We work at the University of Iowa, an academic medical center in the Midwestern United States that implemented an ambient scribe to reduce documentation burden and address provider burnout (Nabla). The tool was introduced in spring 2024 to a limited group of clinicians for a pilot study, then made available enterprise-wide in September 2024. Our qualitative study of the early implementation of the ambient scribe used ethnographic field methods, which included semistructured interviews, participation in weekly team meetings, and attendance at trainings. Our procedures were determined to be nonhuman subjects research by the University of Iowa Institutional Review Board (IRB # 202408209). This manuscript describes findings from the semistructured interviews. We report our methods according to the COREQ Checklist31 criteria.
Study team
The study team included medical informatics officers (E.C., C.I., L.K., J.B., J.M.), ethnographers and implementation scientists (H.S.R., J.V.T.), an additional implementation scientist (K.K.), a researcher specializing in work-life balance (J.L.), and a medical student studying bioethics (N.A.). The ethnographers (H.S.R., J.V.T.) led the interviews; 2 of the medical informatics officers (E.C., C.I.) each coled a proportion of the interviews. The interviews with pilot participants lasted from September to October 2024. Interviews with participants in the enterprise rollout lasted from December 2024 to February 2025.
Participant selection
We developed 2 samples, including a convenience sample of pilot participants and a purposive sample of participants who were part of the enterprise rollout of the ambient scribe. The pilot included a survey13 sent to all pilot participants (n = 38) that had a question about their willingness to participate in a qualitative interview. The clinic champion (J.M.) emailed those who agreed by survey (n = 10); the ethnographers (H.S.R., J.V.T.) reached out to and interviewed those who agreed by email (n = 8). For the broader sample, the clinic champion (J.M.), ethnographers (H.S.R., J.V.T.), and 2 medical informatics officers (E.C., C.I.) identified potential participants using information from a prepost survey (preambient scribe use survey and 30-day postambient scribe use survey) that the clinic champion (J.M.) administered during the hospital-wide rollout. We initially looked for clinicians who responded to the preambient scribe survey, the 30-day postambient scribe survey, and said “yes” to being willing to participate in a qualitative interview (n = 78).
Among those 78 clinicians, we focused on those who generated more than 100 notes or less than 49 notes using the ambient scribe in order to maximize differences.32 Within each of those groups, we prioritized clinicians who scored above 3.325 (the threshold for burnout) on the Stanford Professional Fulfillment Index (PFI),33 which was included as part of the survey sent to all users upon initial login to the ambient scribe. Finally, we also made note of clinicians’ self-report of spending more than 8 h on notes per week (prior to the ambient scribe). The ethnographers (J.V.T., H.S.R.) emailed 28 potential participants; of those invited, 16 accepted. Size for the hospital-wide sample was guided by information power34 and meaning saturation.35 We sought heterogeneity across specialty, documented note volume (>100 vs 0-49), burnout (PFI ≥3.325 vs <3.325), and prior time-in-notes, anticipating meaning saturation with ∼20-25 interviews given focused aims and relatively homogeneous phenomenon. The total sample size was 24 participants (8 from the pilot, 16 from the hospital-wide rollout). Participant characteristics are included in Table 1.
Table 1.
Interview participant characteristics (n=24).
| Pilot | ||||||
|---|---|---|---|---|---|---|
| Number of clinicians | Number by specialty | Number by degree | ||||
| 8 |
|
|
||||
| Enterprise rollout | ||||||
|---|---|---|---|---|---|---|
| Number of notes | Number of clinicians | Number by specialty | Number by degree | Burnout score | Time on notes/week | Age |
| >100 notes | 8 |
|
|
|
|
|
| 0-49 notes | 8 |
|
|
|
|
|
Setting and documentation context
Interviews spanned outpatient ambulatory clinics (family medicine, pediatrics, internal medicine, psychiatry, otolaryngology, oncology, obstetrics/gynecology) and inpatient services (internal medicine, selected surgical/procedural services). Participants discussed ambient scribe output for outpatient clinic notes, new-patient/consult notes, and procedure notes (eg, radiation oncology, gastroenterology), as well as elements of History of Present Illness (HPI), Physical Exam (PE), and Assessment & Plan (A&P).
Data collection
The ethnographers (H.S.R., J.V.T.) led the development of the semistructured interview guide, in conversation with 3 medical informatics officers (E.C., C.I., J.M.) and the medical student studying bioethics (N.A.). The guide for both pilot and hospital-wide participants prompted interviewees to discuss how they used and implemented the ambient scribe in their clinical practice, the learning curve associated with this ambient scribe, and the overall advantages and challenges of using the ambient scribe. We used a naturalistic, ethnographic approach, with an interview guide informed by broad implementation considerations (eg, adoption, utilization, barriers, and facilitators); the complete guide is provided in the Supplementary Material.
Interviewers obtained verbal consent prior to starting the interview, which were recorded and transcribed using Microsoft Teams. The interviews lasted 26-60 min (median: 40 min). Throughout data collection, we did some member checking36 with clinician study team members who were also ambient scribe users. Two medical informatics officers (including 1 Associate Chief Health Information Officers [ACHIO] and 1 Physician Informatics Officer [PIO]) coled a subset of interviews to contribute role-specific context. To mitigate potential power dynamics, we (1) reiterated confidentiality and that data would be deidentified before team discussion and (2) ensured that none of the interviewers were in the participants’ supervisory chains. We note the possibility of social desirability bias despite these steps.
Data analysis
We used thematic analysis37 with an inductive orientation. Initial collaborative coding38 of 2 transcripts by an ethnographer (J.V.T.), the 2 medical informatics officers who coled some interviews (E.C., C.I.), and the clinic champion (J.M.) produced a provisional codebook. While inductive, our reading was informed by sensitizing concepts from implementation science (eg, intervention characteristics, inner setting, characteristics of individuals)39 and documentation “genre” (see below) as lenses to notice patterns without imposing a priori themes. The primary ethnographer then coded all transcripts, iteratively refining codes and memos; themes were developed through team dialogue40 rather than consensus reliability metrics. The ethnographer (J.V.T.) also conducted interim, more targeted analyses that supported local presentations by the clinic champion (J.M.) to check evolving interpretations of the qualitative findings. An additional member of the study team (J.L.) independently coded the transcripts to verify and validate our codes, definitions, and themes.
Our codebook included 5 categories of codes, including codes related to clinician experience, documentation, interactions with patients, implementation, and features of the technology (eg, dot phrases). Using MAXQDA,41 a qualitative data analysis software, to manage the coding process, the most common co-occurring codes were Benefits to Using The Ambient Scribe and Impact on the Patient Encounter, followed by Challenges to Using The Ambient Scribe and Note Elements. Definitions of those codes are included in Table 2. Analytically, we attended to perceptions of “voice” in AI-assisted notes, how documentation signals role/identity and expertise, and how clinicians write for different audiences (eg, supervisors, referring clinicians, patients, coders). Identity signaling and communication became cross-cutting constructs during theme development. The idea of documentation “genres”—treating note sections (eg, HPI, A&P) as patterned forms that serve community-recognized purposes—sensitized us to the ways in which clinicians signal identity, expertise, and audience in text and how AI-assisted drafting may alter these practices. Evidence of clinician perceptions of voice, as well as the lens of genre, guided theme interpretation.
Table 2.
Codes used in thematic analysis.
| Code | Code definition |
|---|---|
| Benefits to Using The Ambient Scribe | Benefits for users. Examples: “saves time,” “more face time with patients” |
| Impact on the Patient Encounter | Discussions of clinician experience of patient encounter, including improved relationships with patients, as well as focusing more on the patient and less on the computer. Includes statements about clinicians feeling more present and engaged, as well as making eye contact with patients |
| Challenges to Using The Ambient Scribe | Challenges for users. Examples: “The ambient scribe can’t do this” or “The ambient scribe isn’t great at that” |
| Note Elements | Discussions of the sections of the note, including Subjective/History of Present Illness, Assessment & Plan, Physical Exam. Includes statements like, “Here’s what it did for the subjective…” and discussions about what the ambient scribe created in those sections. Also includes qualitative descriptions of what the ambient scribe puts in different sections of the note, such as its “too long,” “too short,” or what the ambient scribe left in and left out |
Results
We spoke with 24 clinicians representing a range of specialties (Table 1). Overall, we spoke with 7 Family Medicine clinicians, 5 Internal Medicine clinicians, 4 Pediatricians, and then individual clinicians representing Emergency Medicine, OBGYN, Oncology, Otolaryngology, Radiation Oncology, Psychiatry, and Rheumatology. In terms of how many notes they created, we spoke with 8 clinicians who made more than 100 notes using the ambient scribe, as well as 8 clinicians who made less than 49 notes (we did not count how many notes the pilot participants made). In terms of level of training, we spoke with 20 MDs and 4 ARNP/PAs. Overall, we found familiar benefits (eg, better patient engagement, higher job satisfaction, and improved efficiency), as well as some ambivalence (eg, too much or too little detail, unfamiliar or inadequate formatting, and loss of their voice). Participants in inpatient and procedure-heavy settings more often described reliance on existing templates and team-standardized formats, reporting less perceived benefit from the ambient scribe for highly templated procedure or discharge notes. Ambulatory clinicians more frequently emphasized gains in feeling present with patients and conversational flow during routine clinic visits. Consistent with the genre lens, we report how clinicians described genre-specific benefits and frictions (eg, longer HPIs; inadequate A&P) alongside broader implementation experiences.
Clinicians expressed feeling more focus in interactions with patients
Our qualitative results on clinicians’ perspectives on the benefits of the ambient scribe suggest that they perceive that using ambient scribes creates more opportunities to focus on the patient, more confidence in interactions with the patient in general, and greater job satisfaction. Many clinicians noted how they made more eye contact with patients during clinic visits. Clinicians connected eye contact with feeling more present and engaged. The ability to have more eye contact with the patient was reported to improve the ease in which clinicians could “hear” what the patient was saying. One clinician described how, when not using the ambient scribe, “I’m supposed to be listening, but I don’t hear everything” (Participant 21). They went on to describe how, “[I] get lost as I’m trying to like look them [patients] in the eye, look at the computer, launch this, launch that, check result” (Participant 21). Another clinician reflected about using the ambient scribe,
When I get to make eye contact, well, that’s a positive reinforcer. I was like, wow, I really enjoyed that conversation [with the patient]. And I didn’t have to spend it typing. And then these few minutes of rest in my brain while I’m confident that [the ambient scribe] is at least going to remind me of all the important details. (Participant 11)
More eye contact, and being more able to listen, yielded more satisfaction in clinician interactions with patients. One clinician described how the ambient scribe helped them interact with patients in ways they always wanted to. They said, “the more present you can be, I think the better. It’s always something I’ve aimed for, but I do think I am better at it since I’ve started using this” (Participant 16).
Clinicians used documentation to personalize their practice and demonstrate their expertise
Within reflections about the benefits of using the ambient scribe for their clinical work, clinicians also described ambivalence about using the ambient scribe for their documentation work. Clinicians were careful about balancing the work they had put in to crafting their personal note style with the potential benefits of using the ambient scribe. Once clinician reflected,
I have practiced and tried writing OB notes a gazillion different ways over the course of the last, like, 14 years. And I feel like my flow makes the most sense for me right now. And so to give that flow up, I feel like would be—, I don’t know, very uncomfortable. But I’m not opposed to trying. (Participant 6)
Clinicians that worked in inpatient settings, or in procedure-heavy specialties (eg, radiation oncology, gastroenterology), often had personalized workflows to incorporate note templates used by other clinicians on their teams, or they had adopted workflows to facilitate the documentation practices already established by their colleagues. One resident physician said, “I would love to have it for procedures, but we already have our template set up and I just feel it’ll probably be faster if I just do [it that way]” (Participant 20). Another resident further explained that, on the inpatient side, they found the ambient scribe “to be a little more burdensome… in part because of some of the pre-templated formats that [they] can use and, in part based on how the Epic workflow is already organized” (Participant 17). They reflected on how these pretemplated formats and existing workflows were “already very streamlined into making [documentation work] really efficient” (Participant 17).
In addition to existing efficiencies, younger clinicians, in addition to physicians still in training, often spoke about how they had to write notes in a specific way, to satisfy the expectations of their supervising clinicians. One clinician characterized the assessment and plan as the “crown jewel” of the note when teaching medical students, telling the students, “that’s why you went to school, that’s why you went to training, is to come up with an assessment and a plan” (Participant 18). More seasoned clinicians still reported sentiments such as, “I like to sound smart in my notes” (Participant 1). Characteristics of the note, including format and word choice, signaled both deference to authority, and competence in their own work. It was not always clear how to use the ambient scribe in a way that maintained that signal. Clinicians talked about how some note sections could be either too detailed or not detailed enough. Specifically, the HPIs often had “too much non-pertinent detail” (Participant 15). The A&P regularly had too little information, meaning not what the clinician wanted include and/or “not always exactly as [the clinician] would have put it” (Participant 21).
Using this technology meant clinicians shared their voice with the ambient scribe
Prior to our implementation of the ambient scribe, many clinicians had spent a long time crafting their note templates, and, as 1 clinician said, “for everybody, I think the biggest barrier is how do I get it to look and sound the way I want it to” (Participant 17). In using the ambient scribe, clinicians expressed needing to get comfortable with notes that did not sound like them. At the same time, clinicians were willing to share their voice with the ambient scribe because they noticed how using the ambient scribe was more efficient and improved their relationship with patients. One clinician explained how,
It’s not exactly like the words I would choose to write in the note, but it has the same exact meaning, and I think that’s been more important to me than, like, making it my own, is that, like, I’m—, it’s still conveying the exact same information that I would convey, and it really, I think, has improved that relationship with the patient, right? I just feel more present with them. (Participant 14)
Another clinician wrestled with homogenizing their voice and the ambient scribe’s voice. They explained,
[The ambient scribe] is not my voice, right? [The ambient scribe] is a very different voice from my voice. My notes have a certain voice … And so when I read the note, it doesn’t sound like me. But in the interest of efficiency, I’m OK. It’s fine. I mean, let’s get the note done. (Participant 18)
That same clinician further explained,
A lot of what I’m trying to get across in my assessment is, like, either reassurance, you know, to the referring provider or concern, or something like—, I’m trying to communicate something without just exposing I am very worried or, you know, ‘do not worry about this’ because that just doesn’t come across right. So there is a voice component to that … like the nuance of the assessment, and where I’m trying to communicate my overall sentiment in a way … It’s how you how you phrase things. You know, how you list out kind of the elements that you’re paying attention to, and when the follow up’s gonna be, you know there’s kind of stuff goes into that that I think the referring provider gets. … [the ambient scribe] tends to do things very bluntly, which I understand, right. It gets in there and just bluntly, it’s fine. But it’s just, you know, yeah, it lacks some of that nuance. (Participant 18)
In the interest of efficiency, to “get the note done,” this clinician was willing to use the ambient scribe for their documentation work. However, in the interest of communicating their nuanced assessment as a specialist to a referring provider, like a hospitalist or primary care doctor, this clinician resisted using the ambient scribe. Other clinicians shared this awareness that either the ambient scribe could not recreate their voice, or something important was lost when their voice was obscured. Practically, this meant that the ambient scribe seemed to shift clinicians’ relationship to notes; 1 clinician described the shift as being less of a “content creator” and “more content editing” (Participant 16).
Discussion
Overall, the ambient scribe supported clinicians in ways that allowed them to focus more on their patients and feel more efficient in their documentation work. At the same time, using the ambient scribe highlighted how clinicians used their documentation to do more than move patients through the healthcare system. Clinicians talked about using their documentation to personalize their practice, demonstrate their expertise, manage other clinicians’ impressions of them, and communicate sensitive clinical findings. Other researchers have discussed similar findings, dichotomizing “structured notes” from notes that provide space for “expressivity.”42,43 Some of this work has been pulled forward in discussions of the quality of notes produced by ambient scribes,6,22 but more recent work to develop an evaluation framework emphasized “fluency, completeness, and factuality,”44 and does not mention concepts like “expressivity.” In our findings, the ambient scribe sometimes provided more challenges than benefits when trying to reincorporate nuance, evidence of expertise, and communication patterns for conveying concern that other clinicians would recognize. In incorporating the ambient scribe into these aspects of their documentation, clinicians started doing a different kind of work, reincorporating their voice into the ambient scribe output.
One way to theorize this new kind of work is using “genre theory” from archival studies, specifically as it has been taken up in sociocultural analyses as “genre tracing.”28,45–53 A “genre” is defined as a “pattern of communication that conforms to community norms” in order to “realize particular communicative and collaborative purposes.”46,53 Each of the elements of a clinical note, the Subjective/HPI, the Physical Exam, and the A&P, are each a genre. Using the concept of a genre to characterize the elements of the clinical note allows us to notice and describe the kinds of activities that are central to documentation work, but are not reflected in the current discourse about the impact of ambient scribes on documentation work. Clinicians’ discussions of how they personalize their practice, demonstrate their expertise, manage other clinicians’ impressions of them, and communicate sensitive clinical findings are not captured by current measurements of efficiency and job satisfaction.
Using the data we collected, and guided by genre theory, we can offer a brief discussion of 2 potential avenues of future inquiry. First, it will be important to pay attention to how ambient scribe technology impacts social memory.54 A genre “represents [a] community’s history of problem solving; its solutions are preserved in its genres.”52 One question to ask is, how important are individual clinical voices to the record of those solutions? We heard clinicians say “it’s not exactly like the words I would choose to write in the note,” and “it doesn’t sound like me.” What does a history of problem-solving look like if it is written by one homogenized voice? In addition to representing a community’s history of problem solving, genres are also “traditions of producing, using, and interpreting artifacts.”52 Here, several pieces of data from our study might resonate, including the perspectives of the trainees, who spoke about writing notes in a specific way to satisfy their supervisors, and the perspective of the more senior clinician, who claimed that one of central purposes of medical training was to learn how to write an assessment and plan. Learning to write a clinical note is an ongoing process that socializes early career clinicians into the profession, not only as a way of doing but also as a way of thinking. Where and how will that socialization work happen if the ambient scribe writes, or helps to write the clinical note? At what stage of training should ambient scribes be introduced?
Second, it will be important to pay attention to discussions of accountability. People use genres to “account” for themselves; they “draw from values and social principles to demonstrate the coherence and rationality of their behavior.”28,49 Clinicians do this by templating a note or organizing a workflow, some of which might be to standardize their behavior, and some of which might be to standardize the behavior of others. Is the homogenization that ambient scribes seem inclined to produce the same as the standardization that clinicians often seek out? In genre theory, “giving and asking for accounts” is a form of identity work.51 Yakel deftly illustrates this by describing the history of how radiologists standardized their “radiological report,” and how by creating a standard to which they held themselves, they more easily demonstrated their specific expertise, and thus their contribution to the community of clinicians.51 Will ambient scribes obscure this identity work and make it harder to claim a place in the community of clinicians, by homogenizing the note without connecting this homogenization to any kind of standard that reflects an expertise?
Strengths and weaknesses
This study has several strengths, including its ethnographic methodology and rigor. At present, there is only 1 study55 reporting qualitative study of ambient scribes; our ethnographic methodology extends the conversation by providing not only a method (ie, qualitative) but also a way of thinking56 (ie, with genre theory) about the sociotechnical contexts in which clinicians use ambient scribes. Moreover, the rigor of our approach is multifaceted: (1) on reliability: the composition of the study team reflects the composition of the sample (ie, clinicians from Family Medicine, Internal Medicine, and Pediatrics), (2) on validity: independent coding by a study team member (J.L.) who was not part of data collection, and (3) on transparency: we report our Methods using the COREQ. At the same time, our study has several limitations. In a subset of interviews, informatics officers (including 1 ACHIO and 1 PIO) were present. Although we implemented safeguards (confidentiality assurances; no interviewers were in participants’ supervisory chains), some participants may have tempered criticism. The sample size, while adequate for a qualitative study, may mean that our findings have limited generalizability; at the same time, our findings resonate with published quantitative work and confirm published qualitative work. Also, we did interview a wide range of users (eg, a person who created 366 notes and a person who created 2 notes), we did not interview nonusers, who may have unique perspectives on the ambient scribe. Our findings are based on only 1 institution, which is an academic medical center, and may not represent clinical workload seen at nonacademic medical centers. Finally, our findings are based on only 1 ambient scribe, which may have different features and outputs than others.
Conclusion
In early enterprise implementation, clinicians reported feeling more present with patients and perceived efficiency improvements, alongside new work reconciling AI-drafted text with local documentation practices. Implementation approaches attentive to documentation genre and audience may help align documentation with clinical communication needs.
Clinical implications
Our findings suggest 2 practical next steps for implementation: (1) local, role-specific “genre packs”—role- and audience-specific prompts or templates for HPI and A&P, consults, admissions, and progress notes and (2) onboarding models that pair novices with superusers and offer short office-hours to calibrate length, accuracy, and tone.
Pragmatic lessons
Pilot vs rollout: Pilot participants more frequently cited rapid iteration with the vendor representative and clinic champion (eg, prompt refinements, quick feedback loops) and clearer onboarding touchpoints. Hospital-wide rollout participants described more heterogeneous onboarding experiences and greater reliance on peers. Several challenges reported early (overlong HPIs; sparse A&P) were later mitigated for some users via local prompt/“dot phrase” guidance from peers.
Pragmatic lessons for adopters include: (1) brief, role-specific prompt exemplars by note genre (HPI vs A&P) lessen the need for postediting; (2) early “office-hours” improves uptake; (3) pairing novices with local “superusers” accelerates calibration; (4) explicitly addressing when not to use the tool (eg, complex procedures or highly standardized templates) curbs frustration.
Key take-aways
Familiar benefits
Better patient engagement
Higher job satisfaction
Improved efficiency
Some ambivalence
Too much, or too little detail, sometimes in the wrong place in the note
Formatting sometimes felt unfamiliar, inadequate, or incorrect
Note was not in their voice
Implications of using AI scribes
Ability for more focus on patients, and more confidence in interaction with patients
May impact diversity of voices present in histories of clinical problem-solving
Raises questions about medical training and the socialization of trainees
Risks potential homogenization of notes and obscuring clinical expertise
Future directions
Supplementary Material
Acknowledgments
The authors would like to acknowledge colleagues with whom ongoing collaboration has created an environment of intellectual collegiality and friendship. These individuals include Hilary Mosher and Pete Taber (through the CHART Lab project), the AI Workgroup at the University of Iowa, including Anjali Narain, Colleen Campbell, Jeydith Gutierrez, Laura Seegmiller, and Lauris Kaldjian, and finally a group of Family Medicine clinicians and researchers focused on AI-related work, including Aaron Kunz, Emily Welder, Julie Armin, Kelly Skelly, Michael Maharry, and Yinghui Xu. The authors would also like to acknowledge the generosity of the clinicians who participated in the interviews on which we report in this paper.
Contributor Information
Jen Van Tiem, Department of Family and Community Medicine, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, United States; Implementation Science Center, University of Iowa, Iowa City, IA 52242, United States.
Elizabeth Cramer, Department of Family and Community Medicine, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, United States.
Christopher Iverson, Department of Internal Medicine, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, United States.
Korey Kennelty, Department of Family and Community Medicine, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, United States; Department of Pharmacy Practice and Science, College of Pharmacy, University of Iowa, Iowa City, IA 52242, United States.
Noah Andrys, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, United States.
Julie Lee, Department of Health Management and Policy, College of Public Health, University of Iowa, Iowa City, IA 52242, United States.
Lindsey Knake, Stead Family Department of Pediatrics, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, United States.
Jason Misurac, Stead Family Department of Pediatrics, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, United States.
James Blum, Department of Anesthesia, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, United States.
Heather Schacht Reisinger, Implementation Science Center, University of Iowa, Iowa City, IA 52242, United States; Department of Internal Medicine, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, United States.
Author contributions
Jen Van Tiem (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Writing—original draft, Writing—review & editing), Elizabeth Cramer (Formal analysis, Investigation, Writing—review & editing), Christopher Iverson (Formal analysis, Investigation, Writing—review & editing), Korey Kennelty (Resources, Writing—review & editing), Noah Andrys (Writing—review & editing), Julie Lee (Validation), Jason M. Misurac (Conceptualization, Formal analysis, Writing—review & editing), Lindsey Knake (Conceptualization, Writing—review & editing), James M. Blum (Conceptualization, Resources, Supervision), and Heather Reisinger (Conceptualization, Formal analysis, Investigation, Methodology, Resources, Supervision, Writing—original draft, Writing—review & editing)
Supplementary material
Supplementary material is available at Journal of the American Medical Informatics Association online.
Funding
This study received no direct funding, but received institutional support from Dr Jeffrey Quinlan, the Department Executive Officer of Family and Community Medicine at the University of Iowa Carver College of Medicine. H.S.R.’s work was supported by the University of Iowa’s Year 2 P3 Strategic Initiatives Program through funding received for the project entitled “High Impact Hiring Initiative” a program to strategically recruit and retain talented faculty.
Conflicts of interest
All authors declare no financial or nonfinancial competing interests.
Data availability
Data are provided within the published article. The complete qualitative datasets generated and/or analyzed during the current study are not publicly available due to the data containing information that could compromise research participant privacy/consent but are available from the corresponding author on reasonable request.
References
- 1. Blaseg E, Huffstetler A. Artificial intelligence scribes shape health care delivery. Am Fam Physician. 2025;111:304-305. [PubMed] [Google Scholar]
- 2. Zarefsky M; American Medical Association. . Ambient scribes and AI feel like magic in health care. Accessed June 11, 2025. https://www.ama-assn.org/practice-management/digital-health/ambient-scribes-and-ai-feel-magic-health-care
- 3. Stokel-Walker C. The “ambient scribe” tools listening to and summarising your doctor-patient consultations. BMJ. 2025;389:r663. [DOI] [PubMed] [Google Scholar]
- 4. Albrecht M, Shanks D, Shah T, et al. Enhancing clinical documentation with ambient artificial intelligence: a quality improvement survey assessing clinician perspectives on work burden, burnout, and job satisfaction. JAMIA Open. 2024;8:ooaf013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Barak-Corren Y, Wolf R, Rozenblum R, et al. Harnessing the power of generative AI for clinical summaries: perspectives from emergency physicians. Ann Emerg Med. 2024;84:128-138. [DOI] [PubMed] [Google Scholar]
- 6. van Buchem MM, Kant IMJ, King L, Kazmaier J, Steyerberg EW, Bauer MP. Impact of a digital scribe system on clinical documentation time and quality: usability study. JMIR AI. 2024;3:e60020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Bundy H, Gerhart J, Baek S, et al. Can the administrative loads of physicians be alleviated by AI-facilitated clinical documentation? J Gen Intern Med. 2024;39:2995-3000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Duggan MJ, Gervase J, Schoenbaum A, et al. Clinician experiences with ambient scribe technology to assist with documentation burden and efficiency. JAMA Netw Open. 2025;8:e2460637. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Evans K, Papinniemi A, Ploderer B, et al. Impact of using an AI scribe on clinical documentation and clinician-patient interactions in allied health private practice: perspectives of clinicians and patients. Musculoskelet Sci Pract. 2025;78:103333. [DOI] [PubMed] [Google Scholar]
- 10. Haberle T, Cleveland C, Snow GL, et al. The impact of nuance DAX ambient listening AI documentation: a cohort study. J Am Med Inform Assoc. 2024;31:975-979. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Hassan H, Zipursky AR, Rabbani N, et al. Special topic on burnout: clinical implementation of artificial intelligence scribes in healthcare: a systematic review. Appl Clin Inform. 2025;16:1121-1135. 10.1055/a-2597-2017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Lee C, Britto S, Diwan K. Evaluating the impact of artificial intelligence (AI) on clinical documentation efficiency and accuracy across clinical settings: a scoping review. Cureus. 2024;16:e73994. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Misurac J, Knake LA, Blum JM. The effect of ambient artificial intelligence notes on provider burnout. Appl Clin Inform. 2025;16:252-258. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Moryousef J, Nadesan P, Uy M, Matti D, Guo Y. Assessing the efficacy and clinical utility of artificial intelligence scribes in urology. Urology. 2025;196:12-17. [DOI] [PubMed] [Google Scholar]
- 15. Nguyen OT, Turner K, Charles D, et al. Implementing digital scribes to reduce electronic health record documentation burden among cancer care clinicians: a mixed-methods pilot study. JCO Clin Cancer Inform. 2023;7:e2200166. [DOI] [PubMed] [Google Scholar]
- 16. Owens LM, Wilda JJ, Hahn PY, Koehler T, Fletcher JJ. The association between use of ambient voice technology documentation during primary care patient encounters, documentation burden, and provider burnout. Fam Pract. 2024;41:86-91. [DOI] [PubMed] [Google Scholar]
- 17. Rotenstein L, Melnick ER, Iannaccone C, et al. Virtual scribes and physician time spent on electronic health records. JAMA Netw Open. 2024;7:e2413140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Seth P, Carretas R, Rudzicz F. The utility and implications of ambient scribes in primary care. JMIR AI. 2024;3:e57673. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Stults CD, Deng S, Martinez MC, et al. Evaluation of an ambient artificial intelligence documentation platform for clinicians. JAMA Netw Open. 2025;8:e258614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Tierney AA, Gayre G, Hoberman B, et al. Ambient artificial intelligence scribes to alleviate the burden of clinical documentation. NEJM Catal. 2024;5:CAT-23:1–15. [Google Scholar]
- 21. Tierney AA, Gayre G, Hoberman B, et al. Ambient artificial intelligence scribes: learnings after 1 year and over 2.5 million uses. NEJM Catal. 2025;6:CAT-25:1–17. [Google Scholar]
- 22. Balloch J, Sridharan S, Oldham G, et al. Use of an ambient artificial intelligence tool to improve quality of clinical documentation. Future Healthc J. 2024;11:100157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Bracken A, Reilly C, Feeley A, Sheehan E, Merghani K, Feeley I. Artificial intelligence (AI)—powered documentation systems in healthcare: a systematic review. J Med Syst. 2025;49:28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Rabbani N, Ray M, Verhagen E, Hatoun J, Patane L, Vernacchio L. Special topic burnout: ambient artificial intelligence scribes in pediatric primary care: a mixed methods study. Appl Clin Inform. 2025;16:1578. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Tracey P, Garcia P, Punzalan R. Recordkeeping, logistics, and translation: a study of homeless services systems as infrastructure. Arch Sci (Dordr). 2023;23:1-27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Buckley P, Wang Y, Gopalan P. Artificial intelligence scribes in psychiatry. Focus (Am Psychiatr Publ). 2025;23:44-48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Ong JHW, Tung JYM, Sng GGR, et al. A pilot study using ambient artificial intelligence scribes in clinical documentation in an urology outpatient clinic. Published online ahead of print June 5, 2025. BJU Int. 2025;136:415-417. [DOI] [PubMed] [Google Scholar]
- 28. Trace CB. Documenting work and working documents: perspectives from workplace studies, CSCW, and genre studies. 2011. Accessed June 9, 2025. https://ieeexplore.ieee.org/abstract/document/5718605
- 29. Heath C, Luff P. Technology in Action. Cambridge University Press; 2000. [Google Scholar]
- 30. Garfinkel H. Studies in Ethnomethodology. Polity; 2018. [Google Scholar]
- 31. Tong A, Sainsbury P, Craig J. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups. Int J Qual Health Care. 2007;19:349-357. [DOI] [PubMed] [Google Scholar]
- 32. Palinkas LA, Horwitz SM, Green CA, Wisdom JP, Duan N, Hoagwood K. Purposeful sampling for qualitative data collection and analysis in mixed method implementation research. Adm Policy Ment Health. 2015;42:533-544. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Trockel M, Bohman B, Lesure E, et al. A brief instrument to assess both burnout and professional fulfillment in physicians: reliability and validity, including correlation with self-reported medical errors, in a sample of resident and practicing physicians. Acad Psychiatry. 2018;42:11-24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Malterud K, Siersma VD, Guassora AD. Sample size in qualitative interview studies: guided by information power. Qual Health Res. 2016;26:1753-1760. [DOI] [PubMed] [Google Scholar]
- 35. Hennink MM, Kaiser BN, Marconi VC. Code saturation versus meaning saturation: how many interviews are enough? Qual Health Res. 2017;27:591-608. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Birt L, Scott S, Cavers D, Campbell C, Walter F. Member checking: a tool to enhance trustworthiness or merely a nod to validation? Qual Health Res. 2016;26:1802-1811. [DOI] [PubMed] [Google Scholar]
- 37. Clarke V, Braun V. Thematic analysis. In: Teo T, ed. Encyclopedia of Critical Psychology. Springer; 2014:1947-1952. [Google Scholar]
- 38. Bradley EH, Curry LA, Devers KJ. Qualitative data analysis for health services research: developing taxonomy, themes, and theory. Health Serv Res. 2007;42:1758-1772. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Damschroder LJ, Reardon CM, Widerquist MAO, Lowery J. The updated consolidated framework for implementation research based on user feedback. Implement Sci. 2022;17:75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Goedken CC, Livorsi DJ, Sauder M, et al. “The role as a champion is to not only monitor but to speak out and to educate”: the contradictory roles of hand hygiene champions. Implement Sci. 2019;14:110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. VERBI Software. MAXQDA 2022 [computer software]. VERBI Software; 2021. maxqda.com [Google Scholar]
- 42. Ebbers T, Kool RB, Smeele LE, et al. The impact of structured and standardized documentation on documentation quality; a multicenter, retrospective study. J Med Syst. 2022;46:46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Rosenbloom ST, Denny JC, Xu H, Lorenzi N, Stead WW, Johnson KB. Data from clinical notes: a perspective on the tension between structure and flexible documentation. J Am Med Inform Assoc. 2011;18:181-186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Wang H, Yang R, Alwakeel M, et al. An evaluation framework for ambient digital scribing tools in clinical applications. NPJ Digit Med. 2025;;8:358-313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Bazerman C. The orders of documents, the orders of activity, and the orders of information. Arch Sci. 2012;;12:377-388. [Google Scholar]
- 46. McKenzie PJ, Davies E. Genre systems and “keeping track” in everyday life. Arch Sci. 2012;12:437-460. [Google Scholar]
- 47. Trace CB. What is recorded is never simply ‘what happened’: record keeping in modern organizational culture. Arch Sci. 2002;2:137-159. [Google Scholar]
- 48. Shankar K. Order from chaos: the poetics and pragmatics of scientific recordkeeping. J Am Soc Inf Sci. 2007;58:1457-1466. [Google Scholar]
- 49. Trace CB. Ethnomethodology: foundational insights on the nature and meaning of documents in everyday life. J Doc. 2016;;72:47-64. [Google Scholar]
- 50. Andersen J. The concept of genre in information studies. Annu Rev Info Sci Technol. 2008;42:339-367. [Google Scholar]
- 51. Yakel E. The social construction of accountability: radiologists and their record-keeping practices. Inf Soc. 2001;17:233-245. [Google Scholar]
- 52. Spinuzzi C. Tracing Genres through Organizations: A Sociocultural Approach to Information Design. The MIT Press; 2003 [Google Scholar]
- 53. Yates J, Orlikowski WJ. Genres of organizational communication: a structurational approach to studying communication and media. Acad Manage Rev. 1992;17:299-326. [Google Scholar]
- 54. French SA. What is social memory? South Cult. 1995;2:9-18. [Google Scholar]
- 55. Shah SJ, Crowell T, Jeong Y, et al. Physician perspectives on ambient AI scribes. JAMA Netw Open. 2025;8:e251904. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56. Van Tiem JM, Schacht Reisinger H, Friberg JE, et al. The STS case study: an analysis method for longitudinal qualitative research for implementation science. BMC Med Res Methodol. 2021;21:27. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data are provided within the published article. The complete qualitative datasets generated and/or analyzed during the current study are not publicly available due to the data containing information that could compromise research participant privacy/consent but are available from the corresponding author on reasonable request.