In a November 6 interview accompanying an episode of The UMB Pulse, researcher Katherine E. Goodman describes what happens when an AI scribe listens to a doctor’s appointment. The tool drafts a note from the conversation. A clinician then reviews the draft before signing it for the medical record.
How the draft reaches the clinician
Goodman describes a typical workflow in which an app records the conversation, turns the audio into a transcript and uses generative AI to summarize it. The physician receives the draft, can edit it and signs the final note. The workflow gives the clinician a chance to catch mistakes, including details in the transcript that the draft left out.
The interview describes possible benefits such as allowing a clinician to pay more attention to the conversation during the visit. Goodman also emphasizes that systems vary in how they handle both the recorded material and the resulting draft.
Why missing details matter
Goodman draws attention to clinically significant omissions. If a tool leaves a detail out, the draft can look coherent and still lose useful information. In her example, the draft could keep a report of leg pain but omit the patient’s recent long flight. The point of the example is the difference between what the patient said and what a later reader would see in the note.
What to ask about the recording
The interview suggests asking what is captured, how long it is kept, who holds it and what it is used for. Goodman distinguishes the audio recording from the transcript because products can handle them differently. She also describes variation in how the choice to use a scribe is explained before or during an appointment. A patient trying to understand their own visit needs the process for that clinic and tool.
