Articles

Ambient artificial intelligence (AI) scribes are on the rise for physicians across the nation. Experts estimate this rapidly deploying technology is already in use by more than 30% of physicians - and growing by the day. The solution addresses key pain points: reducing the burden of documentation, replacing expensive manual labor, and preserving the most precious asset in a medical practice – the physician’s time. This technological advancement comes at a crucial moment, as physicians increasingly seek sustainable ways to protect work-life balance.

Operating behind the scenes in the exam room, this HIPAA-compliant technology listens to conversations between clinicians and patients, generating structured clinical notes in real time. Through natural language processing, the system learns the physician’s voice and style, allowing documentation to read as if it were personally authored. Like traditional transcription tools, these structured notes are uploaded directly to the patient’s medical record.

For many physicians, ambient AI scribes feel transformative. Yet, whether already in use or under consideration, the technology warrants thoughtful evaluation. As with earlier generations of speech recognition tools, challenges remain - and awareness is essential.

Glitches. Like any technology, system errors can occur. When clinicians rely fully on the tool, disruptions may lead to documentation gaps, particularly if issues go unnoticed for extended periods. Periodic quick checks to confirm the solution is functioning properly can help mitigate problems.

Hallucinations. So-called AI hallucinations can pose serious patient safety risks. These fabricated details may be difficult to detect without careful review.

Omissions. Critical elements of the encounter - such as symptoms, findings, or the plan of care - may be excluded entirely from the documentation, making post-encounter proofreading essential.

Relevancy. Lacking true clinical reasoning, the system may include irrelevant data or misrepresent the significance of key information, reinforcing the importance of a thoughtful review after the visit.

Misinterpretation. Because ambient AI relies solely on audio input, it cannot capture nonverbal cues. Context and nuance may be lost, increasing the risk of misunderstanding. Taking a moment to add commentary on body language or other nonverbal cues after the encounter can improve note accuracy.

Attributions. With multiple voices present during a visit, statements may be incorrectly assigned. Research has also identified disparities tied to language and ethnicity, increasing vulnerability to error. Consider the tool much like a new staff member - take the time to train it before deployment and monitor it carefully as it gets off the ground.

In practice, ambient AI scribes are best viewed as powerful assistants - not autonomous replacements for clinical judgment. With appropriate oversight, review processes, and governance, they can meaningfully reduce administrative burden while maintaining the integrity and safety of your patient records.

Telemedicine is Back

The newly-enacted law, the Consolidated Appropriations Act 2026, extended key pandemic-era Medicare telemedicine flexibilities through December 31, 2027. Under the law, Medicare patients may continue to receive telemedicine services from their homes, with no geographic restrictions on the originating site.

The legislation preserves several critical flexibilities, including the waiver of geographic site limitations, the expanded list of eligible Medicare providers who may furnish telemedicine services, and the ability for Federally Qualified Health Centers (FQHCs) and Rural Health Clinics (RHCs) to serve as distant-site providers for non-behavioral telemedicine care. It also suspends initial and annual in-person visit requirements for Medicare behavioral and mental health telemedicine services.

In addition, the law permits audio-only telemedicine for a broad range of Medicare services. Audio-only communication may also be used on a permanent basis for home-based telemedicine when the clinician is capable of video technology, but the patient is unable or unwilling to use it.