For months, barely a week has passed without a blogger, discussion group, or main-stream media asking if neural machine translation (NMT), with its artificial intelligence (AI) technology, will replace human translators. NMT is the new media darling. Explore this article’s convenient table that compares NMT’s features with other MT technologies so you can decide if it’s ready for prime time.
It started with a Google research report in September 2016. Mind you, Google was not the first to experiment with or deploy an NMT system. Both Facebook and Microsoft had production NMT systems running before Google. Google’s NMT report stated,
In some cases human and GNMT translations are nearly indistinguishable on the relatively simplistic and isolated sentences sampled from Wikipedia and news articles for this experiment(1).arXiv:1609.08144v1 [cs.CL] 26 Sep 201
That sounds really balanced. Nonetheless, reports of Google’s report shakes the market. Why is there so much hype and controversy?
In my opinion, the media needs the ratings. They engaged in what is now popularly called “fake news.” They ignored Google’s 18 pages preceding my quote above, They splashed the headline, “Nearly Indistinguishable From Human Translation”— Google Claims Breakthrough(2). Note the use of the words nearly indistinguishable, the quotes, and the twisted word order. Search for the headline’s quoted text in the article and it just isn’t there. The Slator.com quote is “fake news.”
Kirti Vashee, a popular translation technology blogger, quickly identified this discrepancy and posted The Google Neural Machine Translation Marketing Deception(3). Regrettably, not even Kirti’s well-balanced and reasoned logic could forestall the flood of speculation and hype that ensued.
Our Market’s Imagination
When was the last time I witnessed a new translation technology capture our market’s imagination?
In 2006. The University of Edinburgh released the Moses Toolkit, an open source software project for statistical machine translation research. The market sprang to life. New technology companies vied to leverage the free software for profit. Within 3 years, a dozen or more companies popped up with cloud-based services offering “near human quality” translations that simply needed post-editing (4)(5), thus lowering the cost by replacing human translators with less-skilled translators. When SMT was the reigning media darling, Jaap van der Meer, an industry pundit, predicted this SMT Utopia in his 03 Nov 2009 article, Let a Thousand MT Systems Bloom(6):
Looking into the future, I see a thousand MT systems blooming. I see fortune for the translation industry, and new solutions to overcome failed translations. I see a better world due to improved communications among the world’s seven billion citizens. And the reason why I am so optimistic is that the process of data effectiveness is joining hands with the trend towards profit of sharing.Jaap van der Meer, an industry pundit, 03 Nov 2009
Where are those dozen companies now? Some have passed or morphed. Some are still around. Two have opened A-B testing to customers between tried-n-true SMT systems and new NMT systems (7)(8). KantanMT’s founder Tony O’Dowd doesn’t know if NMT is any better. Reading between the lines in his interview, these A-B tests present the appearance that KantanMT are up-to-speed with current trends, capture some attention with the newest media darling, and could justify (or not) a new NMT technology investment before realizing an ROI on SMT.
The hyperbolic predictions aside, this is only the tip of the iceberg. The parallels run deep between NMT’s emergence in the SMT era, and SMT’s emergence in the now-bygone RbMT era (rules-based machine translation technology). I created this scorecard to help us track the technologies.
= possible but weak
= stronger than 1
= stronger than
CAT = responsibility of the CAT or user application, not the MT technology
N/A = not applicable with today’s technology
= doesn’t support this feature
= might work, but no proven use cases
better for User Generated Content and broad domain material such as patents
documentation and even software
post-editing and durable changes
on-the-fly translations of short-shelf-life content
retains corrections to terminology
use the most likely term, not the one you expect
predictable but the sentences may be awkward
unpredictable but sentences are more fluid
fast update, maintain (daily or more frequently)
longer update cycles (once or twice a year)
expensive to license
can be free open source
heavy on linguistic resources
heavy on processing resources (see hardware below)
makes more fluid sentences
can handle bad grammar
significantly better with controlled authoring
choice for minority languages
match for languages like French and Spanish
performs better for Japanese, German, Russian, Korean
language pairs out of the box
ready to customize for your domain and preferred terminology
hardware resources required
modern notebook PC
time to create new engine
< 1 hour
Special thanks to Lori Thicke at Lexworks because I based the scorecard on her 2012 blog, 13 Differences between SMT and RBMT that You Need to Know (9). I updated Lori’s list based on our experience through 2017 and comments Alon Lavie (senior manager of machine translation at Amazon) and Chris Wendt (machine translation program manager at Microsoft) during a recent podcast interview on Globally Speaking Radio (10).
(1) https://arxiv.org/pdf/1609.08144v1.pdf, bottom of page 18