“Its speech-recognition team swapped out part of their old system for a neural network and encountered, in pretty much one fell swoop, the best quality improvements anyone had seen in 20 years.”
On December 14th, 2016, the New York Times Magazine published a long article on Google’s new translation tools based on Artificial Intelligence or AI. The tool was stealthily introduced without any hoopla in August and the improvement over the often laughably bad machine translation out there (including Google’s) was huge, in fact it’s very close in quality to human translation. The article, in my opinion, may be the best long form piece I’ve seen on complex technology and is worth the hour or so of your time it takes to read and absorb.
So, why am I writing about this on a blog dedicated to content management? One of the things the easyDITA CCMS does well is improve efficiencies in translation workflows. Because the content in a CCMS is structured at a granular level, it is possible to limit redundant translations of content that has already been translated, saving huge amounts of time and money. Translation technology is a core part of our business value and software architecture.
Disruption Can’t Be Ignored
In the translation and localization business, Google’s technology, along with others being developed by Baidu, Facebook, and others, is causing an upheaval that may make translation by humans one of the first businesses to be significantly disrupted by AI. Disruption, by its nature, cannot be dismissed or denied. This is not going to happen, it is happening and it is incumbent on us, as professional communicators, to understand the implications.
For many of our customers, improved translation and localization workflows have helped them globalize their businesses and organizations. It is complex, costly, and the potential for errors and omissions is very real. But the tradeoff is access to entirely new markets. The promise of AI, in this example, is a great reduction in the cost and time required to open up new languages, introduce new products, and capture new revenue. AI won’t replace human translators, but it is likely to change their roles away from actual translation and towards editing and curating those translations to ensure they are accurate.
The Disruption Isn’t Limited To Translation
As the article notes in its epilogue, many professions like radiology and case law can be improved by AI, with the equivalent disruptive effects. In our field, keeping up on these developments is essential to maintaining and improving our communication and content management architectures, so they remain cutting edge. I highly recommend taking the time to read this article.