Notes produced by AI can be organized by medical condition, recommendations, and next steps, and are not verbatim transcripts, with the technology capable of distinguishing voices and filtering out irrelevant chatter.

The technology can distinguish voices and organize notes by sections, while avoiding a verbatim transcript and focusing on relevant clinical content.

Jefferson plans to extend ambient listening tools to emergency departments and nurse hand-offs, while Penn has piloted an in-house scribe tool used by about 1,200 providers in ambulatory and outpatient settings.

Ethicists warn of risks including potential errors, accent misinterpretation, privacy and consent concerns, and data retention questions as AI scribes expand use in clinical settings.

Clinicians note benefits such as improved note quality, better eye contact with patients, and the ability to narrate the exam for the scribe to capture relevant details.

Regulatory oversight remains limited, with the FDA not currently regulating AI as medical devices in the same way, though its digital health committee is exploring potential future regulation.

Hospitals advocate a human-in-the-loop model, ensuring clinicians review AI-generated notes before finalizing them for the medical record.

Penn Medicine studies show AI scribes can reduce in-visit EHR time by about 20% and post-visit documentation by about 30%, improving patient interaction.

Ambient listening and AI scribe tools are being used in exam rooms to record conversations and generate structured medical notes, enabling clinicians to engage more fully with patients and reduce documentation time.

Clinicians report that AI notes are often more comprehensive and useful than manual notes, but require careful review and approval to ensure accuracy, with a human-in-the-loop approach emphasized.

Regulatory and ethical considerations are highlighted, including patient consent and secure handling of sensitive information as adoption proceeds.

Health systems stress patient consent, data security, and that AI recordings should not replace human judgment, reinforcing a “human in the loop” stance.

Jefferson Health and Penn Medicine have rolled out AI scribes that summarize visits and outline medical issues and next steps, with notes appearing within minutes after encounters and requiring clinician approval.

Initial deployments in these systems include in-house or pilot AI solutions that create notes within minutes post-visit, reviewed and approved by clinicians before becoming part of the record.