An AI scribe is not
dictation

Dictation and an ambient AI scribe both involve a clinician speaking near a microphone, which is why the two get treated as interchangeable. They are not. Dictation takes a clinician's own spoken words as its input: a summary, a letter, a set of findings, spoken deliberately and usually close to the final structure. The output is a transcript of that speech, produced by voice recognition software or a medical typist, then typed or pasted into the record. An ambient scribe takes a different input altogether: the natural, unstructured conversation between clinician and patient during the consult, captured with the patient's consent. The output is not a transcript of that conversation. It is a structured clinical document, a progress note, a referrer or GP letter, a discharge summary, drafted from what was actually said.

That distinction sets up everything else. Dictation only ever produces what the clinician chooses to say out loud. An ambient scribe drafts from what happened in the room, including detail the clinician never got around to narrating because narrating it out loud was never the point of the conversation.

Dictation puts a real cognitive task onto the clinician's plate: composing the note. Before a word can be spoken, the clinician has to decide what belongs in the record, put it in a defensible clinical order, and turn it into sentences. That composition work is separate from the clinical reasoning that happened during the consult itself. Some clinicians do it in the room, breaking eye contact with the patient to speak findings into a handset. Others hold the whole consult in their head and dictate a batch of notes at the end of a session or the end of the day.

Either way, the spoken output still has to land somewhere. Voice recognition software converts it to text directly, or a medical typist transcribes it and someone checks the transcription, before it goes into Best Practice, MedicalDirector, Genie, Cliniko or Halaxy. Dictation removes the physical typing. It does not remove the work of deciding what to say and holding it in memory until it is said.

An ambient scribe removes the narration step rather than the typing step. It listens through the consult itself and drafts the note from the conversation that was already happening: the history the patient gave, the examination findings discussed out loud, the plan explained back to the patient. The clinician is not composing a second, separate account of the consult purely for the record. The clinical conversation and the source material for the note are the same event.

The clinician's job moves from authoring a note from a blank page or from memory, to reviewing and correcting a draft against their own clinical judgement. That is still real work, and it is why a named clinician reviewing and signing before anything becomes part of the record matters as much for an ambient scribe as it does for a dictated note. What changes is where the effort goes: composing from nothing, versus editing something already drafted from the actual encounter.

Dictation asks a clinician to handle the same clinical information twice: once while treating the patient, and again while composing and speaking a structured account of it for the record. That second pass is genuine mental work, and it is why dictation backlogs build up, notes get finished at the end of the day when detail has faded, or on a weekend when the clinician would rather not be doing paperwork at all.

An ambient scribe collapses that to a single pass. The clinician conducts the consult; the drafting happens from that same conversation, not from a second recollection of it afterwards. The saving is not primarily typing speed. It is the removal of a whole separate composition task that used to sit between seeing the patient and finishing the note.

The Medicare Benefits Schedule requires a contemporaneous, clinically relevant record of a consult, and AHPRA's expectations for practitioners point the same way: notes finished close to the encounter, while detail is accurate and complete. A dictation backlog works against that directly. If a note is composed from memory two days after the consult, it is no longer being written at the point of care, and quality drifts as recall fades.

Because an ambient scribe drafts from the consult as it happens, the note is ready for review immediately afterwards rather than queued for an end of day dictation session. That makes a same day, contemporaneous record easier to sustain without adding unpaid hours after the last patient leaves. The clinician still has to read the draft and sign it before it becomes a record; the scribe drafting sooner does not change who is accountable for what the record says.

Some things hold regardless of whether a note started as dictation or as an ambient draft, including how the underlying audio and text are handled; the practicalities of that are covered separately in how patient information is stored and secured.

• Patient consent is different in kind, not just degree: agreeing to a dictated summary is a different conversation from agreeing to have the whole consult listened to, and practices should treat it that way rather than folding it into standard paperwork.
• The treating clinician remains the author of record either way. How a note was drafted does not change who is accountable for its accuracy once it is signed.
• Health information is sensitive information under the Privacy Act 1988 (Cth) and the Australian Privacy Principles, so encryption, storage location and retention matter whichever method produced the note.
• The practice's existing system of record, whether that is Best Practice, MedicalDirector, Genie, Cliniko or Halaxy, is still where the note has to end up. Neither dictation nor an ambient scribe replaces it; both feed into it.