In a groundbreaking new study spearheaded by the University of Otago, Wellington, researchers have uncovered widespread adoption among New Zealand general practitioners (GPs) of artificial intelligence-driven scribes to transcribe patient consultations. This swift integration of AI scribes in primary care reflects a broader trend toward digital innovation within healthcare, yet it also illuminates pressing challenges related to legal oversight, ethical considerations, data security, patient consent, and doctor-patient dynamics that remain unresolved.
The study, which surveyed 197 healthcare professionals across various roles in primary care, including nurses, nurse practitioners, and rural emergency care providers, provides an incisive snapshot of how these AI transcription tools are reshaping clinical workflows. Conducted in early 2024, this exploratory survey revealed that approximately 40% of respondents are actively employing AI scribes, signifying a notable infiltration of this technology despite the absence of comprehensive regulatory frameworks governing their use.
Users of AI scribes report a spectrum of experiences that blend optimism with skepticism. A substantial portion of clinicians found these tools beneficial for optimizing record-keeping efficiency. Nearly half of the AI scribe users estimated that integrating the technology into every consultation could save between 30 minutes to two hours daily, showcasing the potential for significant time-saving benefits. However, this optimistic outlook is tempered by concerns over the necessity to invest considerable time into editing and correcting errors generated by the AI—a factor that sometimes nullifies these time gains.
.adsslot_kgwHZsifYa{width:728px !important;height:90px !important;}
@media(max-width:1199px){ .adsslot_kgwHZsifYa{width:468px !important;height:60px !important;}
}
@media(max-width:767px){ .adsslot_kgwHZsifYa{width:320px !important;height:50px !important;}
}
ADVERTISEMENT
The primary source of contention lies in the variability of transcription accuracy. Several practitioners expressed frustration with omissions and errors in AI-generated clinical notes, undermining trust in the tool’s reliability. Notably, instances of “hallucinations,” a term used to describe AI’s tendency to generate plausible yet incorrect information, were cited as a deterrent to unreserved reliance. Additionally, the system’s deficiencies in interpreting local linguistic nuances, including diverse New Zealand accents and te reo Māori, further complicate its usability.
The impact of AI scribes extends beyond transcription accuracy into the dynamics of the consultation itself. Over half of those surveyed noted that the inclusion of AI scribes altered their interaction with patients, compelling clinicians to verbalize physical examination findings and clinical reasoning explicitly to ensure precise capture by the system. This behavioral adjustment marks a significant departure from traditional consultation styles, potentially influencing the quality and fluidity of doctor-patient communication.
Despite these challenges, some healthcare providers observed that the use of AI scribes paradoxically enhanced patient engagement by allowing more sustained eye contact and active listening. This suggests that when effectively integrated, AI transcription tools could recalibrate the clinician’s focus towards empathetic patient interaction, an element often compromised due to administrative burdens. The nuanced relationship between technology and human connection in clinical practice warrants thorough longitudinal studies.
Ethical and legal concerns loom large in this evolving landscape. The study highlights uncertainty regarding compliance with New Zealand’s existing professional regulations and data protection laws. Clinicians shoulder the responsibility for ensuring the accuracy and confidentiality of clinical notes, irrespective of AI involvement. Professor Angela Ballantyne, the lead researcher and bioethics expert, emphasizes the critical need for health practitioners to meticulously review AI-generated documentation, though this requirement often offsets the anticipated efficiency gains.
Data security emerges as another pivotal concern. Many AI scribe platforms rely on cloud-based infrastructures established by international technology companies. This arrangement raises questions about data sovereignty, access controls, and vulnerability to cyberattacks. Of particular sensitivity within the New Zealand context is the management of Māori data, which invokes principles of indigenous data sovereignty and demands culturally informed governance models that respect tribal autonomy and data custodianship.
Adding momentum to this technological shift, the National Artificial Intelligence and Algorithm Expert Advisory Group (NAIAEAG) at Health New Zealand has recently endorsed two ambient AI scribe solutions—Heidi Health and iMedX—for clinical deployment. This endorsement reflects growing institutional recognition of AI’s potential to augment healthcare delivery, while underscoring the necessity of scrupulous evaluation of privacy, security, and ethical dimensions.
Yet, Patient autonomy remains a cornerstone of any bid to integrate AI tools into healthcare. Professor Ballantyne warns against presuming implicit patient consent for AI involvement. The survey revealed that only 59% of practitioners regularly sought explicit consent from patients before using AI transcription services, revealing a gap in current practice. As AI tools become more embedded within clinical operations, robust policies are essential to guarantee patients’ right to opt out without compromising access to quality care.
Looking ahead, forthcoming guidance from the Medical Council of New Zealand is anticipated to formalize standards for AI utilization in health settings, likely mandating explicit patient consent and outlining best practice frameworks. The convergence of regulatory oversight, clinician education, and technological refinement is poised to shape the trajectory of AI scribes in the years to come, balancing innovation with safeguarding patient rights.
Technological advancements continue at a rapid pace, and AI transcription tools are becoming increasingly sophisticated. Improvements in natural language processing, contextual understanding, and accent recognition promise to mitigate current shortcomings. When complemented by comprehensive training for health professionals and stringent governance mechanisms, AI scribes hold substantial promise to revolutionize clinical documentation, reducing administrative workloads while enhancing the quality of care delivery.
Despite the hurdles, the University of Otago’s study provides an optimistic appraisal of AI scribes, suggesting a future where these tools are seamlessly integrated into primary healthcare. By fostering collaboration between technologists, clinicians, ethicists, and policymakers, the healthcare system can harness AI’s transformative potential and navigate its associated risks effectively.
This study, published in the Journal of Primary Health Care, opens critical discourse on the intersection of artificial intelligence and medical practice, inviting ongoing research and dialogue in this fast-evolving field.
Subject of Research: People
Article Title: Using AI scribes in New Zealand primary care consultations: an exploratory survey
News Publication Date: 8-Aug-2025
Web References: http://dx.doi.org/10.1071/HC25079
Image Credits: University of Otago
Keywords: Artificial Intelligence, AI Scribes, Primary Care, Patient Consent, Data Security, New Zealand Healthcare, Clinical Documentation, Bioethics, Māori Data Sovereignty, Doctor-Patient Relationship
Tags: adoption of AI scribesAI in healthcareAI technology integration in clinical workflowsdigital innovation in primary careefficiency of AI transcription toolsethical considerations in AIhealthcare professionals’ perspectives on AIimpact of AI on doctor-patient relationshipslegal challenges of AI useNew Zealand healthcare study on AI usepatient consent and data securityregulatory frameworks for AI in medicine
