Jobs declining
Speech recognition has made first-pass transcription abundant; value is moving to verification, specialist terminology, and accountable records.
Last updated · 2026-07-23
Job Transition Score
Job Transition Scores are an editorial analytical framework — estimates for comparison, not scientific measurements.
What the job was
Transcriptionists listened to recorded speech and produced accurate written records, often applying speaker labels, timestamps, templates, and specialised terminology.
Task breakdown
Generating words from clear audio is largely machine-capable. The difficult work is deciding what was actually said in noisy, accented, overlapping, technical, or legally sensitive material.
| Task | Human | AI | Future |
|---|---|---|---|
| Produce a first-pass transcript | Listens, pauses, rewinds, and types. | Speech-to-text models generate a timed draft quickly. | Machine drafts become the default starting point. |
| Identify speakers | Uses voices and conversation context. | Diarisation separates and labels recurring voices. | Humans correct overlaps, unnamed participants, and identity-sensitive labels. |
| Resolve specialist terms | Applies medical, legal, or technical vocabulary. | Uses custom dictionaries and contextual language models. | Domain-qualified reviewers verify terms that affect decisions. |
| Edit for readability | Removes false starts and applies style rules. | Reformats verbatim speech into clean notes or summaries. | Editors choose the appropriate level of fidelity for the use case. |
| Certify the record | Checks against audio and accepts responsibility. | Highlights uncertain spans and possible inconsistencies. | Human certification remains where records support care, litigation, or compliance. |
Drivers
Routine meetings, interviews, and media production adopt automated transcription quickly. Courts, healthcare, and regulated records retain stricter review because a fluent transcript can still be materially wrong.
What shrinks
What remains human
Rising value
Emerging form
The occupation narrows toward transcript quality assurance, accessibility editing, and specialist record review rather than general-purpose typing.
Entry level
General transcription offers less defensible entry-level work. A stronger path combines language accuracy with a regulated domain, accessibility practice, media production, or records management.
Career transition map
Outlook
These horizons are editorial estimates. Adoption speed varies by industry, regulation, trust, and cost.
Most workflows start with automated text, while humans correct important or difficult material.
General transcription demand contracts; specialist verification and multilingual review remain.
Human work is concentrated in certification, accessibility, archival, and high-ambiguity contexts.
Connected map