postAGI
← All occupations

Medical transcriptionist

9 sourcesUpdated 2026-11-06Sector: Health

BLS counts 42,000 medical transcriptionists in 2025, down from 95,100 in 2010, and blames speech recognition. Most of the fall came before generative AI. No source separates AI's share.

Partially replaced
Where this job stands
  1. Not yet
  2. Augmented
  3. Partially replaced (current status)
  4. Replaced

The marker shows the status level, not a measured percentage.

Tasks automated
Unknown
Year replaced
Unknown
Confidence
Medium. Several independent sources point the same way, but each covers only a slice of the occupation, or a gap remains.

AI tools doing this job

Task breakdown

TaskStatusToolSource
Turning dictation into a first-draft report (speech recognition software)Partially replacedSpeech recognition softwarelink
Reviewing and editing software-generated draftsNot yet-link
How we rate →

Evidence

The BLS Occupational Outlook Handbook shows a long, steady fall. Taken from successive editions of the same page: 95,100 jobs in 2010 (capture from January 2013, which then projected 6% growth), 70,000 in 2014, 58,500 in 2019, 49,000 in 2022 and 42,000 in 2025 (current edition, last modified October 29, 2026, projecting a 4% decline, or 1,900 jobs, by 2035). The editions use different projection bases, so the trend is approximate. On those figures, about 46,000 of the 53,000 jobs lost between 2010 and 2025 were gone by 2022, before generative AI tools spread. BLS already blamed technology in 2013, when it said back-end speech recognition was making each transcriptionist far more productive and limiting employment growth. The 2017 and 2021 editions added outsourcing abroad, the 2024 edition named speech recognition and electronic health record software, and the current edition names speech recognition and natural language processing. It says they allow physicians to document encounters in real time, reducing the need for medical transcriptionists, and that they raise productivity, allowing more transcription by fewer workers. The current page does not use the words AI or outsourcing. BLS still lists reviewing and editing drafts prepared by speech recognition software as a duty, and projects about 6,200 openings a year, all from workers who leave.

The evidence on ambient AI scribes concerns clinicians, not transcriptionists, and none of it measures transcriptionist employment. The Peterson Health Technology Institute (March 2025) says ambient scribes are poised to become one of the fastest technology adoptions in healthcare history, and presents them as a scalable alternative to human scribes and talk-to-text dictation. A JAMA Network Open study (October 2025; 263 clinicians in 6 health systems) found burnout fell from 51.9% to 38.8% after 30 days, and says nothing about transcriptionists. The professional body AHDI's 2025 to 2028 strategic plan (December 2024) mentions AI only as a planned item, saying it includes developing a plan for documentation created using AI as a tool.

This page counts speech recognition as AI, as the methodology page explains. That is why the status is partial: BLS ties a measured fall in demand to those tools, but it does not say what share of the work they do or separate them from outsourcing and electronic health record software. Nothing found meets the bar for replaced, since tens of thousands of jobs remain. Limits: I found no workforce survey of transcriptionists, and a January 2026 report about positions moving to AI in Newfoundland could not be confirmed.

References

Suggest a correction on GitHub →