Real-time ai translation went from a novelty to internal infrastructure
real-time ai translation went from a novelty to internal infrastructure: deloitte counts 47% of multinationals using it for internal communication in 2026, up from 12% in 2023, and 83% of language service providers now sell machine translation with post-editing as a standard product. gemini 2.5 pro topped 16 language pairs at the 2025 wmt benchmark.
the translators being replaced are not the best ones, they are the first ones.
Context
The WMT25 findings say Gemini 2.5 Pro was the best overall system in the general machine translation task, placing in the top cluster for 14 of 15 evaluated language pairs (Machine Translate summary of the findings).
DeepL's 2026 Language AI Report (Mar 10, 2026) surveyed 5,005 business leaders in five countries and says 68 percent still rely on manual processes and traditional tools for translation.
Gemini 2.5 Pro as the top system matches WMT25, but the summary read says 14 of 15 language pairs, not 16. Deloitte's 47 percent of multinationals, the rise from 12 percent in 2023 and the 83 percent of language service providers selling post-editing were not seen in the sources read, so unsupported here, not refuted. The DeepL survey points the other way for enterprise: most leaders still rely on manual translation processes. 'The first ones' replaced is the author's.
Related work
Watch next
- Find the Deloitte survey and the 83 percent language service provider figure.
Sources
- Machine Translate: WMT25machinetranslate.org
- arXiv: preliminary ranking of WMT25 general machine translationarxiv.org
- PR Newswire: DeepL research on manual translation processesprnewswire.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 1 October 2026 at 20:16 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →