Open access peer-reviewed chapter - ONLINE FIRST

Understanding the Human-Computer Interaction of Ambient AI in Healthcare: A Qualitative Study of Early Provider Experiences

Written By

Thomas R. Wojda, Francis Vino Dominic Savio, Harry Hochheiser, Anusha Amaravathi, Krish Bharat and Gary Fischer

Submitted: 28 April 2025 Reviewed: 28 May 2025 Published: 03 July 2025

DOI: 10.5772/intechopen.1011340

Artificial Intelligence in Medicine and Surgery - An Exploration of Current Trends, Potential Opportunities, and Evolving Threats, Volume 3 IntechOpen
Artificial Intelligence in Medicine and Surgery - An Exploration ... Edited by Stanislaw P. Stawicki

From the Edited Volume

Artificial Intelligence in Medicine and Surgery - An Exploration of Current Trends, Potential Opportunities, and Evolving Threats, Volume 3 [Working Title]

Dr. Stanislaw P. Stawicki and M.D. Thomas R. Wojda

Chapter metrics overview

14 Chapter Downloads

View Full Metrics

Abstract

Ambient artificial intelligence (AI) documentation tools offer the potential to reduce clinician burden and enhance workflow efficiency. However, limited research has explored provider experiences with ambient AI, particularly from a human-computer interaction (HCI) perspective. To qualitatively evaluate provider experiences one month after implementation of ambient AI documentation tools across a large integrated health system. This was a qualitative thematic analysis embedded within a broader mixed-methods evaluation. Open-ended survey responses were collected from providers who had adopted ambient AI systems (Abridge or DAX) and analyzed using Braun and Clarke’s reflexive thematic analysis approach. Sentiment analysis and interrater reliability assessment supported the rigor of coding. Among 204 survey respondents, 103 provided free-text feedback. Six major themes emerged: (1) improved efficiency and workflow, (2) enhanced patient interaction, (3) documentation quality and editing burden, (4) learning curve and workflow adaptation, (5) specialty-specific performance and suitability, and (6) provider skepticism and trust in AI. While many providers reported time savings and improved patient engagement, concerns persisted around documentation accuracy, editing workload, specialty-specific challenges, and patient privacy. Sentiment analysis revealed a balanced distribution of positive, negative, and mixed perceptions. Early experiences with ambient AI highlight both its promise and its challenges. Successful integration will require specialty-specific customization, enhanced training, transparent data practices, and sustained efforts to build provider trust. Attention to HCI principles and real-world workflow alignment will be critical for broader adoption.

Keywords

  • ambient artificial intelligence
  • clinical documentation
  • human-computer interaction
  • thematic analysis
  • healthcare technology adoption

1. Introduction

Electronic health records (EHRs), while foundational to modern clinical care, are broadly cited as a key reason for clinician burnout because of administrative and cognitive demands [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]. In response, numerous strategies have been introduced to reduce charting demands, including clinical documentation support staff, team-based charting models, voice recognition, and automatic speech recognition technologies [11, 12, 13, 14, 15, 16]. However, these interventions often fall short in scalability, cost-effectiveness, or seamless integration into clinical workflows [17].

Ambient artificial intelligence (AI) driven transcription systems, leveraging sophisticated large language models represent a promising technological advancement in clinical documentation. These systems passively listen to provider–patient interactions and generate clinical notes with minimal input from the clinician, offering the potential to improve documentation quality, reduce time spent on charting, and increase professional satisfaction. Recent studies suggest ambient AI scribes may reduce after-hours EHR use, improve workflow efficiency, and enhance provider engagement [18, 19, 20].

Despite growing interest and early signs of efficacy, most research to date on clinical ambient AI has focused on quantifiable outcomes such as documentation time and burnout scores [21, 22, 23, 24]. Far less is known about the qualitative aspects of provider experience [25, 26]—particularly the human-computer interaction (HCI) dynamics that influence adoption, trust, and long-term integration into clinical care. Understanding how clinicians perceive and interact with these systems is essential for responsible deployment and iterative design improvements.

Recent qualitative studies have begun to fill this gap. Bundy et al. interviewed early users of DAX Copilot and found that while many physicians appreciated the reduction in cognitive and administrative burden, concerns persisted about hallucinations, misgendering, and pressure to increase visit volume [25]. Relatedly, Shah et al. conducted in-depth interviews with providers participating in an ambient AI pilot and identified key facilitators and barriers to adoption, emphasizing workload relief, patient engagement, and concerns around note accuracy and system access [26]. Both studies underscore the complexity of provider-AI interaction and the contextual factors that shape acceptance and utility.

This study builds upon and extends that literature by analyzing open-ended responses from clinicians one month after implementation of ambient AI technology across a large integrated health system. This work focuses on thematic insights derived from free-text feedback provided by clinicians who participated in a post-implementation survey. Rather than comparing specific products (e.g., Abridge vs. DAX/Nuance), this analysis centers on the relational, emotional, and workflow-oriented experiences of providers engaging with ambient AI tools early in their adoption.

By applying qualitative thematic analysis to these narratives, our study aims to deepen understanding of the HCI patterns emerging from early ambient AI use in clinical settings. These insights may help healthcare organizations, technology developers, and informatics leaders foster more human-centered, acceptable, and sustainable ambient AI systems in routine practice.

Advertisement

2. Methods

2.1 Study design

This was a qualitative thematic analysis embedded within a larger mixed methods evaluation of ambient AI. The objective of the qualitative component was to examine provider experiences one month after using ambient AI documentation tools and to identify emerging themes that characterize the human-computer interaction between clinicians and technology. This inductive, exploratory approach was selected to generate hypotheses for future study and to capture rich narrative insights that are not easily measured through quantitative methods. The project was reviewed and approved as a Quality Improvement (QI) initiative by the health system and was conducted in alignment with institutional QI guidelines.

2.2 Setting and participants

The study was conducted within the University of Pittsburgh Medical Center (UPMC), a large integrated healthcare and finance delivery system with facilities across Pennsylvania, New York, and Maryland. Providers working in any clinical setting—primary care or specialty—were eligible to participate, provided they completed a pre-implementation survey prior to using ambient AI technology. Eligible participants included physicians and advanced practice providers (APPs).

2.3 Ambient AI implementation context

Providers had access to either of two ambient AI solutions: Abridge or Nuance’s Dragon Ambient eXperience (DAX). The system-wide implementation included a dedicated website with training materials, as well as on-demand technical support and in-person “elbow” support to assist with onboarding. Providers were introduced to the technology via internal communication campaigns in the year leading up to the survey distribution.

2.4 Data collection

One month after implementation, providers who had completed the pre-survey were sent a follow-up survey via Qualtrics. The survey included both structured and open-ended questions. Here we focused exclusively on responses to the following free-text prompt:

“Please share any additional comments on how ambient AI has impacted you.”

Participants could respond anonymously or choose to include their contact information. A reminder email was sent one week after the initial invitation to improve response rates.

2.5 Data analysis

We used Braun and Clarke’s six-phase reflexive thematic analysis framework to analyze the free-text responses: (1) familiarization with the data, (2) generating initial codes, (3) searching for themes, (4) reviewing themes, (5) defining and naming themes, and (6) producing the report [27]. The lead investigator reviewed and coded all 103 responses line-by-line, generating a comprehensive codebook of descriptive and analytic codes. These codes were then grouped into broader axial categories to reflect emerging patterns.

To enhance the rigor and credibility of the coding process, a second physician independently coded a random subset of 12 qualitative responses (approximately 10% of the dataset) using the initial codebook. Each response was assigned a primary thematic category by both coders. Interrater reliability was assessed using Cohen’s Kappa statistic, which yielded a value of κ = 0.80, indicating substantial agreement between coders (z = 6.33, p < 0.001). Discrepancies were resolved through discussion to finalize thematic assignments and support the validity of the coding framework.

Following axial coding, the unit of analysis shifted to individual codes. Codes were systematically organized into thematic categories, with illustrative quotes selected for each. Sentiment analysis was also performed to deepen interpretation.

2.6 Reflexivity and rigor

The lead investigator is a physician with training in clinical informatics, which may influence interpretation of provider feedback in favor of technological optimism. To mitigate potential bias, an independent coder reviewed a subset of responses, and consensus coding was used to finalize theme assignments. An audit trail of codes, category definitions, and thematic evolution was maintained in Excel to ensure transparency and reproducibility.

Advertisement

3. Results

A total of 1586 providers did the pre implementation survey. 204 providers responded to the 1-month post implementation survey (13%). Of those, 103 providers left responses (50%). Individual responses were also categorized based on sentiment: 43 (42%) were positive, 24 (23%) were negative, 33 (32%) were mixed, and 3 (3%) were neutral. Positive responses emphasized improved efficiency and workflow, while negative responses cited transcription inaccuracies and increased editing burden. Mixed responses commonly reflected both benefits and challenges, with providers appreciating the time-saving potential of AI while acknowledging its current limitations. Through thematic analysis, we identified six major themes characterizing provider experiences with AI-assisted documentation (Figure 1).

Figure 1.

Major themes characterizing provider experience using artificial ambient intelligence.

3.1 Theme 1: Efficiency and workflow impact

Providers reported that ambient AI technology improved documentation efficiency and helped reduce after-hours charting. The ability to automatically generate clinical notes allowed clinicians to focus more on patient care during encounters and spend less time writing notes at the end of the day. Respondents noted that AI-assisted transcription helped streamline their workflow, particularly for post-visit documentation.

“Amazing efficiency add — I document after encounters, so it is so much faster afterwards.”

“Overall, with my current usage over the past 3–4 weeks, I have reduced my work beyond office hours by 2–3 hours per day. I anticipate this to improve even further in the future.”

Providers also described a reduction in cognitive load, as they no longer had to mentally retain encounter details for later documentation. Instead, they were able to edit and refine AI-generated notes, which they felt was more efficient than starting from a blank note.

“The mental burden and pressure of remembering everything immediately after the patient visit and writing it down is greatly alleviated.”

However, efficiency gains varied depending on workflow and documentation style. Some clinicians found AI-generated notes helpful only in specific scenarios and reported that any potential time savings were offset by the need for detailed proofreading and corrections.

“It has not been helpful or increased efficiency. In fact, it makes me less efficient since I have to slowly proofread all notes, which is slower than just saying it right the first time myself.”

3.2 Theme 2: Enhanced patient interaction

Another cited benefit of ambient AI technology was its ability to reduce provider distraction during patient encounters, allowing for more engaged, face-to-face communication. Feeling less tethered to the computer enabled more natural conversations and stronger rapport with patients.

“Ambient AI has allowed me to spend more face-to-face time with my patients (rather than face-to-screen), and I feel less rushed to record the patient history during the visit. However, it has not yet saved me a great deal of time in my schedule.”

This improvement in clinical presence provided a renewed sense of satisfaction in being able to focus fully on their patients.

“I’m able to talk to my patients like I should.”

“Has injected a hint of satisfaction/joy in allowing me to focus more fully on the patient.”

Moreover, respondents highlighted that ambient AI also improved how well their documentation captured the true scope of their clinical conversations, especially regarding patient education.

“I spend a lot of time educating patients on their plan of care. I love the A&P section of ambient AI because now I don’t have to type or dictate everything I already said. It reflects how much time I truly spend educating patients, when before I may not have charted as much to accurately reflect that due to time constraints.”

However, not all providers viewed the technology as enhancing patient interaction. Some raised concerns about awkwardness in verbalizing exam findings aloud, misgendering, and difficulty interpreting diverse accents—all of which could disrupt rapport. Others noted that some patients were reluctant to consent to AI use, which introduced tension into the encounter.

“The notes are sort of awkward and the speaking aloud of physical exam findings is also awkward. I just haven’t mastered it fully yet.”

“It adds work. It constantly misgenders patients and has difficulty interpreting the accents of some of our patients. Also, many patients are reluctant for us to use this program and do not consent.”

Overall, while ambient AI was seen as a promising tool for enhancing patient-centered care, its success depended heavily on provider adaptation, documentation accuracy, and patient receptivity.

3.3 Theme 3: Documentation quality and editing burden

While efficiency benefits of ambient AI were acknowledged, concerns about documentation accuracy and the need for extensive editing were expressed. AI-generated notes often contained errors, misinterpretations, or inconsistencies, requiring substantial proofreading and correction before finalizing.

“A lot of inaccurate translation/transcription of the encounter, too much time spent to read and edit.”

“It is excellent at picking up HPI items from the patient and my PE and plan. The diagnoses always do not match unless I specifically dictate them out, and it still takes a lot of time proofreading the final notes for signing. But much faster than typing or clicking and pointing.”

Ambient AI appeared more useful for routine follow-ups than for complex or new encounters. In more nuanced clinical scenarios, AI struggled to accurately capture reasoning or new diagnoses.

“Ambient AI is not at all useful for documenting new diagnoses or medical reasoning. It is helpful for follow-up problem-focused visits with stable patients. It is less helpful for new patients.”

A recurring concern was the misrepresentation of clinical details, including incorrect summaries, attribution errors, and misgendering of patients. Pronoun inaccuracies and awkward phrasing were cited as frequent editing burdens.

“I still have to edit notes. I do feel it is helpful, but sometimes it has changed a female patient to a him, so I have to change all pronouns.”

“I find the transcribed verbiage quaint. I have to proofread it obviously, and change a lot of the ‘quaint’ verbiage.”

This need for constant oversight begged the question whether ambient AI truly saved time. A trade-off appeared between the reduced cognitive burden and increased editing workload—particularly when AI-generated notes required significant revision to meet clinical standards.

“The AI helps ease the mental effort required to recall and document patient visits, but it still requires a lot of proofreading, and that still takes up time.”

Despite these concerns, providers expressed optimism that with ongoing refinement, improved contextual understanding, and customization, ambient AI tools could become more accurate and reduce editing time over the long term.

3.4 Theme 4: Learning curve and workflow adaptation

Providers described a varied learning curve in adopting ambient AI documentation tools. While some clinicians adapted quickly and found the technology intuitive, others encountered challenges related to workflow integration, speech pattern adjustment, and limited training resources.

“Although Ambient is easy to use, there is a learning curve in terms of tailoring your interview technique to get as best a product as possible. Also, I have the sense that the program is learning by my edits on what I am looking for.”

“It took me a while to figure out how to phrase things so the AI captured them correctly. I wish there was more guidance on best practices for dictation.”

Respondents illustrated a trial-and-error process during which they learned to modify their phrasing, slow down their speech, or restructure their encounters to optimize AI-generated notes. Some clinicians developed personalized strategies to support more accurate transcriptions, such as outlining the assessment and plan point by point or emphasizing key exam findings aloud.

“Still learning to use Ambient AI most efficiently. Works well for HPI, OK for A&P, not so well for PE (even when I say my findings out loud).”

“Once I figured out a good workflow, I saw a lot of improvement in how I used AI. It’s not perfect, but it’s getting better.”

Others highlighted limitations in current functionality, particularly around physical exam capture, complex multi-problem visits, and team-based care scenarios like teaching rounds or procedural encounters.

“It seems to be good with the history/subjective, but I struggle with the physical exam... My patients are very complex... I still have to spend a lot of time revising the A&P part of the note.”

“I’m still not sure how to use the technology when I have residents with me. For most of those visits, I just type my addendum.”

Providers also noted that while AI integration was not yet seamless, they expected continued improvement and were motivated to seek further training to maximize its utility.

“I am going to request a personal training session to improve use, but I am so thrilled with it.”

“Does a GREAT job generating notes. Next step is to sync with Epic in such a way that it updates diagnoses/current problems as presently Ambient does not support problem-based charting.”

Despite initial challenges, respondents expressed that AI documentation improved with regular use, leading to time savings, reduced stress, and increased satisfaction.

“Every visit it gets better. I get faster. Taken 20 hr a week of note-writing out. It is almost as good as my own typing, which is good enough for saving my sanity!”

“In providing acute care, I am able to be much more efficient due to integrated ambient AI. I don’t have to open the computer in the exam room at all, and at the end of the day when I can’t remember details, my HPI is already written for me to edit.”

Overall, this theme highlights that successful implementation depends not only on the capabilities of the technology, but also on sufficient training, time for adaptation, and alignment with clinical workflows.

3.5 Theme 5: Specialty-specific performance and suitability

The effectiveness of ambient AI documentation varied across clinical specialties, with some areas benefiting more than others. Psychiatry, oncology, rheumatology, and other specialty care settings encountered limitations in the ability to capture the full depth and nuance of clinical encounters.

“In my specialty, a lot of the conversation is nuanced and patient-driven. The AI sometimes struggles to capture the essence of these discussions, leading to summaries that miss key details.”

“For rheumatology, it is generally useful for the HPI/interval history, but not for much else since our notes are too complicated for it.”

“It helps with the HPI but not very good with medical oncology assessment plans. These are complicated patients.”

In specialties that rely heavily on structured templates, longitudinal assessments, or standardized clinical terminology, AI-generated notes required additional editing due to overgeneralization, factual inaccuracies, or missing clinical reasoning.

“Ambient AI probably works well in a primary care setting, but it does not translate well in the type of template-based office visit that our office uses. It also does not pick up many of the testing medical terms that we require.”

“I have not found it to be helpful in psychiatric notes. It doesn’t include important aspects of the session but will add things that aren’t important, take patients at face value and say they have specific symptoms when the patients don’t actually have that.”

In contrast, some specialties with higher patient volumes and more standardized workflows, such as urgent care, reported that ambient AI improved efficiency and reduced after-hours charting.

“In urgent care, I see a lot of repeat cases where documentation is fairly standard. AI makes it easier to complete notes quickly without much manual effort.”

Specialty-specific adaptations could improve AI usability. Suggestions included tailoring the AI to specialty-specific templates, enhancing terminology recognition, and training models to capture multi-problem visits or longitudinal care plans more effectively.

“I wish there were a way to tweak how the AI structures notes by specialty. It works well for some types of visits, but not as much for others.”

“Family medicine: essentially have to document the same visit twice. AI does not capture all diagnoses discussed... must cut and paste each individually AI-generated A/P to the problem list. Extremely cumbersome.”

Overall, while ambient AI showed promise across multiple specialties, its performance was highly dependent on the clinical context and documentation needs. Providers in specialized fields emphasized the need for further customization, accuracy, and refinement to realize the full benefits of the technology.

3.6 Theme 6: Provider skepticism and trust in AI

While providers found ambient AI documentation beneficial, skepticism was expressed regarding its accuracy, reliability, and long-term role in clinical care. Concerns included over-reliance on AI-generated notes, potential for critical omissions, and errors that could undermine trust in the documentation process.

“I still feel like I have to double-check everything. I worry that if I rely too much on it, I’ll miss something important.”

“Ambient AI can work for single problem/focused visits. In my patients, I address 1–2 acute problems, 4–6 chronic problems + preventive care each visit. When I used Ambient AI, it was putting the statements under the wrong problems.”

Concerns were raised about the trustworthiness of AI-generated content, noting that statements were sometimes misattributed or that the AI misunderstood the speaker, leading to inaccuracies in clinical documentation.

“Sometimes the AI assigns statements to the wrong person in the conversation. I have to be really careful to make sure the documentation reflects what actually happened.”

Unease about how patients perceive AI in clinical visits was voiced. While most patients accepted its use, others were hesitant or declined due to privacy concerns or lack of clarity around data handling and storage.

“I have a fair number of patients who distrust AI and decline. Mostly they want to know ‘where’ does all the verbiage go and how long is it kept once I close the note. I have told them what I have been told—that it is automatically deleted—but frankly, I am not sure I even believe that to be true.”

Despite these reservations, conditional optimism was expressed. Trust in ambient AI could improve with ongoing system refinement, better transparency, and increased user control over how notes are structured and edited.

“Too early to tell; the language the AI uses is too broad/general and often too generic or simple — as if writing a letter to a patient instead of a clinical note directed to other health care providers. At this point, I am spending equal time editing my AI note as it would be saving me, so in essence, a wash... I am hoping this system learns ‘my language’ and molds my notes accordingly going forward. This system does keep me from forgetting things patients have said during the exam.”

Overall, provider skepticism stemmed from concerns about accuracy, trust, patient consent, and documentation integrity. While they acknowledged the promise of ambient AI, they emphasized the need for ongoing improvements in reliability, user agency, and clear communication with patients to support widespread adoption.

To synthesize the themes described above, we developed a summary table highlighting reported strengths, weaknesses, and design suggestions derived from provider feedback (Table 1). This table provides a practical, high-level overview of the key insights and considerations for improving ambient AI integration in clinical practice.

StrengthsWeaknessesDesign Suggestions
Reduced after-hours documentationInaccurate or misattributed transcriptionAI customization and clinical context adaptation
Improved patient engagement during visitsInconsistent documentation quality across specialtiesTailored templates for specialty-specific care
Decreased cognitive load for routine notesHigh editing burden for complex encountersImproved contextual accuracy and support for complex visits
Streamlined workflow with continued useSteep learning curve with limited training supportEnhanced onboarding and best practice toolkits
Effective for standardized visits (e.g., urgent care)Patient reluctance due to privacy concernsTransparency about data handling, consent, and storage

Table 1.

Summary of Strengths, Weaknesses, and Design Recommendations Based on Provider Feedback.

Advertisement

4. Discussion

This study explored provider perspectives on ambient AI documentation tools one month after implementation across a large integrated health system. Through thematic analysis of 103 open-ended survey responses, we identified six major themes reflecting the lived experience of clinicians using ambient AI: improved efficiency and workflow, enhanced patient interaction, documentation accuracy and editing burden, learning curve and workflow adaptation, specialty-specific performance, and trust in AI. These responses revealed a nuanced and often ambivalent relationship between providers and AI tools—marked by enthusiasm for time-saving potential and frustration with documentation limitations.

While recent qualitative studies have begun to examine physician adoption of ambient AI scribes through structured interviews, our study adds to the literature by offering a broad, real-world snapshot of provider sentiment across diverse specialties and care settings.

Despite the potential benefits of ambient AI, concerns about documentation quality, accuracy, clarity, and the time required for editing were expressed. Respondents reported that AI-generated notes often contained errors, misgendered patients, misattributed dialog, or failed to capture clinical reasoning—issues that not only reduced efficiency but also raised concerns about documentation integrity. These findings align with previous work, which found that physicians perceived AI-generated notes as overly verbose, occasionally inaccurate, and in need of substantial editing [20, 26].

In tandem with quality concerns, providers described a nontrivial learning curve when integrating ambient AI into their workflows. Successful use often required modifying speech patterns, structuring conversations more deliberately, and developing personal workarounds—such as dictating physical exam findings or outlining plans verbally. While some respondents adapted quickly and reported improvements over time, others cited insufficient training, limited onboarding support, and unclear best practices, consistent with previous concerns [22, 28].

These themes highlight the importance of ongoing user training, clear documentation protocols, and adaptive AI systems that align with clinical workflows and provider expectations. While current tools focus primarily on transcription, our findings suggest that clinicians value functionality that supports customization, contextual accuracy, and note structuring. Health systems deploying these tools should consider robust onboarding programs, specialty-specific training, and feedback loops that allow providers to shape how the technology evolves in practice.

Providers emphasized that the performance of ambient AI varied widely by clinical specialty, with better alignment in high-volume, protocol-driven settings such as urgent care, and more challenges in specialties requiring nuanced narrative documentation, such as psychiatry, oncology, and rheumatology. These findings reinforce prior results, where specialty-specific concerns emerged around note accuracy, editing burden, and applicability to complex care [26]. Our study adds to this literature by highlighting that template-based documentation workflows, high diagnostic complexity, and the need for clinical nuance may limit ambient AI’s usefulness without specialty-level customization.

Beyond functional performance, varying levels of trust in technology was voiced. The need to closely monitor the accuracy of AI-generated content was stated as were concerns about misrepresentation of patient statements, incorrect clinical details, and potential medico-legal implications. Some providers also described patient reluctance to consent to AI documentation, especially when privacy or data handling were unclear. These themes mirror broader concerns in the literature about AI transparency, accountability, and user control, underscoring the tension between innovation and clinician confidence [23].

Together, these findings suggest that broad adoption of ambient AI requires greater trust-building mechanisms, such as explainable AI outputs, customizable templates, and clear communication around data privacy and consent. Developers should consider building specialty-specific versions of ambient AI that better mirror the logic, structure, and terminology of distinct fields. From a systems perspective, supporting clinician autonomy through editable, user-controlled notes, as well as reinforcing AI as a collaborator—not a replacement—may increase both trust and usability.

4.1 Implications for practice, design, and policy

This study offers practical guidance for health systems and AI developers integrating ambient documentation tools into clinical care (Figure 2). First, implementation should be context-aware: ambient AI performs best in settings with standardized workflows and may require specialty-specific adaptations in fields with more complex documentation needs.

Figure 2.

Guidance for Development and Integration of Ambient AI Technology in Healthcare.

Second, robust training and behavioral support are critical. Successful use depends not only on technical integration but also on helping clinicians adapt their communication and workflows. Onboarding should include visit-type-specific examples, speech guidance, and ongoing troubleshooting resources.

Third, developer attention to usability, transparency, and customization is key. Tools that allow for real-time editing, personalized templates, and feedback mechanisms may improve clinician trust and long-term engagement.

Finally, equity must be prioritized. Several providers reported performance issues with non-English speakers and diverse accents, echoing broader concerns about bias in voice recognition. Proactive testing across populations and safeguards against disparities in documentation quality are essential. Collectively, these findings underscore that successful adoption of ambient AI hinges on human-centered design, equitable implementation, and alignment with real-world clinical practice.

4.2 Strengths and limitations

This study offers several strengths. It draws on a large sample of 103 provider responses across multiple specialties within a real-world, large integrated health system. The use of open-ended survey questions allowed for spontaneous, unfiltered feedback, providing rich insight into the nuanced experiences of frontline users. Our thematic analysis captured both anticipated and emergent themes, contributing to a deeper understanding of human-computer interaction in the context of ambient AI documentation.

However the study has several limitations. First, it was conducted within a single health system, potentially limiting generalizability to other organizational or geographic contexts. Second, the responses reflect experiences only one month after implementation, and thus may not capture long-term adaptation, satisfaction, or disengagement. Third, the qualitative data were drawn from provider self-report via free-text responses, which may underrepresent more extreme views or nuanced use cases that might be better captured through interviews. Finally, our analysis focused exclusively on provider perspectives; future work should incorporate patient viewpoints, especially given the growing role of AI in clinical encounters.

4.3 Future directions and research needs

This study highlights several areas for future inquiry. First, given that our data reflects provider experiences only one month after implementation, longitudinal research is essential to understand how perceptions evolve as clinicians gain proficiency and as ambient AI systems mature. Future work from our team will include analysis of qualitative responses collected at 3-month and 6-month follow-up surveys, allowing for a deeper, time-based exploration of provider adaptation, satisfaction, and perceived value. Tracking engagement and documentation quality across these time points will help determine whether initial challenges resolve, persist, or shift over time.

Second, more research is needed to explore the patient perspective on ambient AI, particularly through qualitative inquiry. While our study focused on provider experiences, understanding how patients perceive ambient documentation tools—especially in terms of trust, comfort, and communication quality—is critical to evaluating their broader impact on the clinical relationship. To date, only limited work has examined this dimension. One recent study by Owens et al. [29]. found that patients generally responded favorably to ambient voice technology, reporting a more personable and focused encounter; however, this study relied on quantitative satisfaction metrics and limited open-label feedback, without deeper qualitative analysis of patient attitudes or concerns.

Future work should incorporate interviews, focus groups, or open-ended survey responses from patients to better capture the relational and ethical implications of AI-mediated documentation. This is particularly important in diverse populations where issues of privacy, consent, and data transparency may shape a patient’s willingness to engage. As ambient AI becomes more embedded in care, a holistic understanding of its influence on both clinician and patient experience will be essential for responsible implementation.

Advertisement

5. Conclusion

Ambient AI documentation technology represents a promising frontier in reducing clinician burden and enhancing the quality of the patient-provider interaction. Our study adds to the growing body of evidence by offering a nuanced, real-world snapshot of how providers experience this technology across settings and specialties. While the potential for increased efficiency and improved engagement is clear, meaningful adoption will require attention to documentation quality, user training, trust-building, and equity. As health systems continue to explore AI-enabled tools, centering human-computer interaction, user feedback, and ethical implementation will be critical to realizing the full promise of ambient AI in clinical practice.

Advertisement

Acknowledgments

The author acknowledges the use of ChatGPT for idea generation and organization of the outline of the manuscript.

References

  1. 1. Shanafelt TD et al. Relationship between clerical burden and characteristics of the electronic environment with physician burnout and professional satisfaction. In: Mayo Clinic Proceedings. Rochester, Minnesota: Elsevier; 2016
  2. 2. Gardner RL et al. Physician stress and burnout: The impact of health information technology. Journal of the American Medical Informatics Association. 2019;26(2):106-114
  3. 3. Apathy NC et al. Documentation dynamics: Note composition, burden, and physician efficiency. Health Services Research. 2023;58(3):674-685
  4. 4. McPeek-Hinz E et al. Clinician burnout associated with sex, clinician type, work culture, and use of electronic health records. JAMA Network Open. 2021;4(4):e215686-e215686
  5. 5. Tajirian T et al. The influence of electronic health record use on physician burnout: Cross-sectional survey. Journal of Medical Internet Research. 2020;22(7):e19274
  6. 6. Gaffney A et al. Medical documentation burden among US office-based physicians in 2019: A national study. JAMA Internal Medicine. 2022;182(5):564-566
  7. 7. Li C et al. Electronic medical record-related burnout in healthcare providers: A scoping review of outcomes and interventions. BMJ Open. 2022;12(8):e060865
  8. 8. Arndt BG et al. Tethered to the EHR: Primary care physician workload assessment using EHR event log data and time-motion observations. The Annals of Family Medicine. 2017;15(5):419-426
  9. 9. Budd J. Burnout related to electronic health record use in primary care. Journal of Primary Care & Community Health. 2023;14:21501319231166921
  10. 10. Downing NL, Bates DW, Longhurst CA. Physician Burnout in the Electronic Health Record Era: Are we Ignoring the Real Cause? Leawood, Kansas: American College of Physicians; 2018. pp. 50-51
  11. 11. Gidwani R et al. Impact of scribes on physician satisfaction, patient satisfaction, and charting efficiency: A randomized controlled trial. The Annals of Family Medicine. 2017;15(5):427-433
  12. 12. Lin S et al. Five years’ experience with a medical scribe fellowship: Shaping future health professions students while addressing provider burnout. Academic Medicine. 2021;96(5):671-679
  13. 13. Apathy NC, Holmgren AJ, Cross DA. Physician EHR time and visit volume following adoption of team-based documentation support. JAMA Internal Medicine. 2024;184(10):1212-1221
  14. 14. Shaw JG et al. Primary care 2.0: A prospective evaluation of a novel model of advanced team care with expanded medical assistant support. The. Annals of Family Medicine. 2021;19(5):411-418
  15. 15. Smith PC et al. Practice transformation under the University of Colorado’s primary care redesign model. The Annals of Family Medicine. 2019;17(Suppl. 1):S24-S32
  16. 16. Johnson M et al. A systematic review of speech recognition technology in health care. BMC Medical Informatics and Decision Making. 2014;14:1-14
  17. 17. Shah LM et al. Effects of medical scribes on patients, physicians, and safety: A scoping review. Knowledge Management & E-Learning. 2021;13(4):559
  18. 18. Tierney AA et al. Ambient artificial intelligence scribes to alleviate the burden of clinical documentation. NEJM Catalyst Innovations in Care Delivery. 2024;5(3):CAT. 23.0404
  19. 19. Cao DY et al. Artificial intelligence-driven digital scribes in clinical documentation: Pilot study assessing the impact on dermatologist workflow and patient encounters. JAAD International. 2024;15:149-151
  20. 20. Haberle T et al. The impact of nuance DAX ambient listening AI documentation: A cohort study. Journal of the American Medical Informatics Association. 2024;31(4):975-979
  21. 21. Ma SP et al. Ambient artificial intelligence scribes: Utilization and impact on documentation time. Journal of the American Medical Informatics Association. 2025;32(2):381-385
  22. 22. van Buchem MM et al. Impact of a digital scribe system on clinical documentation time and quality: Usability study. JMIR AI. 2024;3(1):e60020
  23. 23. Misurac J, Knake LA, Blum JM. The effect of ambient artificial intelligence notes on provider burnout. Applied Clinical Informatics. 2025;16(02):252-258
  24. 24. Shah SJ et al. Ambient artificial intelligence scribes: Physician burnout and perspectives on usability and documentation burden. Journal of the American Medical Informatics Association. 2025;32(2):375-380
  25. 25. Bundy H et al. Can the administrative loads of physicians be alleviated by AI-facilitated clinical documentation? Journal of General Internal Medicine. 2024;39(15):2995-3000
  26. 26. Shah SJ et al. Physician perspectives on ambient AI scribes. JAMA Network Open. 2025;8(3):e251904-e251904
  27. 27. Braun V et al. Doing reflexive thematic analysis. In: Supporting Research in Counselling and Psychotherapy: Qualitative, Quantitative, and Mixed Methods Research. London, United Kingdom: Springer; 2023. pp. 19-38
  28. 28. Seth P, Carretas R, Rudzicz F. The utility and implications of ambient scribes in primary care. JMIR AI. 2024;3(1):e57673
  29. 29. Owens LM et al. Effect of ambient voice technology, natural language processing, and artificial intelligence on the patient–physician relationship. Applied Clinical Informatics. 2024;15(04):660-667

Written By

Thomas R. Wojda, Francis Vino Dominic Savio, Harry Hochheiser, Anusha Amaravathi, Krish Bharat and Gary Fischer

Submitted: 28 April 2025 Reviewed: 28 May 2025 Published: 03 July 2025