A medical scribe's job is to write the record while the clinician looks at the patient. That is close to exactly what ambient AI documentation tools promise: listen to the visit and draft the note. Few jobs line up so directly with one AI use case.
The short answer: AI is putting real pressure on scribe work, but someone still has to check and stand behind the record. There is no US Bureau of Labor Statistics profile for medical scribes. The closest one, medical transcriptionists, is projected to decline 4 percent from 2025 to 2035, and the BLS says speech recognition and natural language processing let physicians document patient encounters in real time, reducing the need for transcriptionists. Projections are estimates, not guarantees.
How exposed documentation work is
The International Labour Organization scores occupations' tasks from 0 (cannot be automated with generative AI) to 1 (could be done without a human operator), and groups them into gradients from 1 to 4, with 4 the most exposed; below them sit minimal exposure and not exposed (ILO Working Paper 140, 2025).
| Occupation (ISCO-08) | ILO category | Mean task score (0 to 1) |
|---|---|---|
| Medical Secretaries | Gradient 3 | 0.53 |
| Medical Records and Health Information Technicians | Gradient 3 | 0.52 |
| Medical Assistants | Gradient 1 | 0.35 |
The ILO index does not list scribes separately. Documentation-heavy health jobs sit in the significant-exposure band, while jobs that mix records with patient care score lower.
What the BLS projects
| Occupation | Projected change 2025 to 2035 | Openings a year |
|---|---|---|
| Medical transcriptionists (closest profile) | Decline 4% | About 6,200 |
The BLS already lists reviewing and editing drafts prepared by speech recognition software among transcriptionists' typical duties, which is where scribe work is heading too.
Scribe tasks AI is taking on
- Drafting visit notes from the conversation.
- Structuring notes into history, exam and plan sections.
- Pulling details from earlier records.
- Drafting orders, referrals and patient instructions for review.
What stays with people
- Checking the note: AI can produce confident but wrong statements, which NIST calls confabulation (NIST AI 600-1). A wrong drug, dose or side in a chart can harm a patient. See checking AI output.
- Catching what was not said aloud: exam findings, context and corrections the recording misses.
- Privacy: recording a visit captures very sensitive personal data. See AI agents and personal data.
- Accountability: the clinician signs the record, and a person reviews it first. See human in the loop and accountability.
How the job is changing
Scribing is moving from writing notes to reviewing and correcting AI drafts. Skills in records systems, medical terminology and careful checking carry over to related roles. This page describes the job market and is not medical advice.
See also will AI replace medical assistants?, will AI replace medical billing and coding?, will AI replace doctors? and will AI take my job?
Pay and how to get in
The US Bureau of Labor Statistics has no separate occupation profile for this job, so there is no BLS pay or entry education figure for it. See how this job compares in AI exposure by pay and AI exposure by education level.
Questions and answers
Will AI replace medical scribes?
AI is taking over much of the drafting, but notes still need a person to check them. The closest BLS profile, medical transcriptionists, is projected to decline 4 percent from 2025 to 2035 because of speech recognition and natural language processing.
Is there BLS data for medical scribes?
No. The BLS has no separate profile for scribes; medical transcriptionists are the closest.
How exposed is clinical documentation to generative AI?
In the ILO 2025 index, medical secretaries score 0.53 and medical records and health information technicians 0.52, both in gradient 3.
What scribe work stays human?
Checking notes for errors, adding what the recording missed, protecting patient privacy and supporting the clinician who signs the record.