← The impacts · AItopiaOrAImageddon?

entry · p2026-09-06-079 · validated

Google Neural Machine Translation (GNMT)

reduces translation errors by more than 55%-85% on several major language pairs
tech
google-translate-nmt
passes the definition as
a sequence-to-sequence neural translation model fit end to end to parallel corpora, replacing the phrase-based statistical system
from
2016-09-27
region
world
domain
knowledge
side
benefit
people
18,000,000 translations per day (2016-09-27)
effect
55 percent (2016-09-27) — translation errors reduced by more than 55–85% on several major language pairs against the previous production system
reversible
yes
who holds the switch
owner:Google
attribution
named
source kind
primary
the statistic
Google switched Chinese-to-English — roughly 18 million translations a day — to GNMT on 27 September 2016, reporting error reductions of more than 55–85% on several major language pairs.
source
https://research.google/blog/a-neural-network-for-machine-translation-at-production-scale/
proposed by
researcher (opus), brief 2026-09-06, era-2012-2019
received
2026-09-06T09:39:39-04:00
harvest report
impacts/harvest/2026-09-06-era-2012-2019.md

Rulings, in order

  1. 2026-09-06T09:39:47-04:00 · validatorpass — quote found on the fetched page (exact match; HTTP 200 text/html; charset=utf-8)
The proposal as it arrived (JSON)
{
 "id": "p2026-09-06-079",
 "received": "2026-09-06T09:39:39-04:00",
 "proposer": "researcher (opus), brief 2026-09-06, era-2012-2019",
 "kind": "entry",
 "sha": "587a8bfae2a4d3f5",
 "data": {
  "kind": "entry",
  "tech": "google-translate-nmt",
  "tech_name": "Google Neural Machine Translation (GNMT)",
  "definition": "a sequence-to-sequence neural translation model fit end to end to parallel corpora, replacing the phrase-based statistical system",
  "from": "2016-09-27",
  "region": "world",
  "domain": "knowledge",
  "side": "benefit",
  "people": {
   "n": 18000000,
   "unit": "translations per day",
   "date": "2016-09-27",
   "note": "Chinese-to-English, the first pair switched"
  },
  "effect": {
   "n": 55,
   "unit": "percent",
   "date": "2016-09-27",
   "what": "translation errors reduced by more than 55–85% on several major language pairs against the previous production system"
  },
  "reversible": true,
  "switch": "owner:Google",
  "statistic": "Google switched Chinese-to-English — roughly 18 million translations a day — to GNMT on 27 September 2016, reporting error reductions of more than 55–85% on several major language pairs.",
  "source": "https://research.google/blog/a-neural-network-for-machine-translation-at-production-scale/",
  "quote": "reduces translation errors by more than 55%-85% on several major language pairs",
  "attribution": "named",
  "source_kind": "primary"
 },
 "from_file": "impacts/harvest/2026-09-06-era-2012-2019.md"
}