International Edition
Latest News
Business

Dr. Mara Gordon Says AI Scribes Save Time but Risk Privacy and Accuracy

In a column for NPR, family physician Dr. Mara Gordon argues that artificial intelligence scribes in medical offices save doctors time and improve eye contact, but introduce significant privacy risks and documentation errors. Medical practices across the country…

Dr. Mara Gordon Says AI Scribes Save Time but Risk Privacy and Accuracy

In a column for NPR, family physician Dr. Mara Gordon argues that artificial intelligence scribes in medical offices save doctors time and improve eye contact, but introduce significant privacy risks and documentation errors.

Medical practices across the country increasingly adopt automated software to record patient visits and draft clinical notes. Gordon writes that she uses a smartphone application in her Camden, New Jersey, primary care practice to transcribe conversations. The technology relies on automatic speech recognition to separate the voices of doctors and patients before a large language model converts the transcript into a specialized medical summary.

Proponents point out that the software reduces administrative burdens. It saves me time, Gordon writes, adding that it allows her to look at patients instead of staring at a computer screen.

Audio files use HIPAA-compliant cloud storage and encryption

Although doctors often use personal smartphones to record appointments, the audio files are not stored locally on the devices. They’re uploaded directly to a HIPAA-compliant cloud where the software transcribes the text before deleting the original audio and notes after a set period.

Federal privacy regulations require software companies to encrypt health data, restrict access, and maintain safeguards against breaches. Despite these protections, Gordon admits she feels unsettled by artificial intelligence listening to private medical conversations. Some patients also worry about data security and potential identification.

Automated transcription tools cause clinical inaccuracies and errors

Clinical inaccuracies remain a persistent hazard when relying on automated transcription tools. Gordon describes a case involving a patient with metastatic cancer who insisted he had no diagnosis. The software failed to capture the clinical nuance, generating a note that stated the patient had been misdiagnosed.

Other common errors include swapped genders, incorrect ages, and flipped negatives.

“The difference between ‘I do have chest pain’ and ‘I do not have chest pain’ can be a matter of life and death,” Gordon says.

Dr. Eileen Kim, an assistant professor of internal medicine at the University of California San Diego, warns that overburdened physicians often lack the time to review every automated entry. Errors slip into the electronic health record and propagate across future clinical notes.

Physicians remain legally responsible for all documentation regardless of software assistance. Gordon reviews every generated note before signing the chart.

Can patients refuse to use an artificial intelligence scribe?

Can patients refuse to use an artificial intelligence scribe?

Patients can opt out of the recording software without issue. Gordon states that if a patient expresses concern, she simply types notes on her computer manually as she did before the technology existed.

Are appointment recordings used for research?

Routine office visit recordings are generally deleted automatically to protect patient privacy and limit liability. However, some researchers call the recordings “digital gold” for studying physician-patient communication, though they are not used for research purposes outside of enrolled clinical studies.

Can patients opt out of artificial intelligence across the entire healthcare system?

Opting out of healthcare artificial intelligence entirely is nearly impossible. Kim explains that machine learning systems operate throughout modern medical infrastructure, powering billing, risk calculators, and early warning systems for conditions like sepsis.

About the author: Marcus Liu - Business Editor

MBA and ex‑B bureau chief specializing in global finance and fintech. Marcus speaks Mandarin, Japanese, and English, and has interviewed CEOs from the Fortune 50 to Y‑Combinator unicorns. Marcus Liu delivers sharp analysis on markets, startups, and corporate strategy for investors and entrepreneurs alike.