We are hurtling towards the Singularity. Machines are taking over and soon an Orwellian dystopia will descend.
In almost every aspect of our lives, artificial intelligence is encroaching. Soon an autonomous car will deliver an autonomous pizza that will analyze my enjoyment, or lack thereof, and report back to Dominos.
Should we fear it? Will some super-intelligence turn against Humanity and trigger a post-human future? Or is that just a Hollywood trope? Personally, I look forward to the day when all cars on the highway are enslaved to the grid. I can get some work done and play ‘Words with Friends’ at 80 MPH.
Across the arc of history, many advances in technology have been greeted with fear. Massive shifts in society were brought by the cotton gin, the steam engine, the aeroplane. But, although robots have revolutionized the assembly-lines, skilled workers still design, build and maintain those machines. Human effort has simply shifted further up the value-chain.
Ten years ago, when Machine Translation (MT) was ramping up, we were fearful of our jobs. We thought Google Translate would be the end of our industry. Who needs a translator when they can just copy and paste into an algorithm? Well, it didn’t quite happen that way. Turns out that the machine still needs a human.
I once heard a client liken MT to a microwave oven. “Sure” he drawled. “It can cook my food. But that doesn’t mean it tastes great.”
The Globalization and Localization Association (GALA), states: “MT tools are often used to translate vast amounts of information involving millions of words that could not possibly be translated the traditional way. The quality of MT output can vary considerably; MT systems require “training” in the desired domain and language pair to increase quality.”
Similarly, MT revolutionized the way we approach languages. But it won’t replace the need for skilled professional translators any time soon.
Organizations now translate more words than ever before, at a far lower cost than just ten years ago. Their clients and customers expect localized and accurate content. Speed is an absolute essential. As fast as you can create content, it needs to be ported out into multiple languages. MT will get you the gist. But not the grist. People can discern the difference.
So how can we get the most out of machine translation? The key is to combine the best aspects of each type of translation: the facilities offered by machine translation, and the professional translator’s professionalism and safety. The formula is easy:
Machine translation + human post edition = high quality translation = saving of costs and time
Now, MT is evolving further: into neural machine translation (NMT). Unlike the previous Rule-based and Statistical methods, NMT can translate complete sentences without splitting them: generating more coherent results. In addition, it uses Artificial Intelligence, so it learns “to learn”, providing more “human” results.
When we are looking for a high-quality professional translation, it is important to know where to go. Machine translation can be very useful but applying a free online machine translation tool to a document will in no way guarantee that it will be the same as if it were translated by a professional. Many organizations look for translation agencies who can provide both human and machine translation, with a robust process for post editing. They’re looking for the cost & time advantages that come from state-of-the-art linguistic technology, while ensuring quality through the post-edit.
Does it mean we should never use machine translation by itself? Of course not. MT has practical applications, so long as it´s a well-behaved, well-taught machine, and we know its limitations. When it comes to understanding a text written in a unfamiliar language, and when it’s not particularly complex, MT can help us by offering the “low-cost” version of the document translation. Having said that, this text will never have the appropriate level for its use and diffusion beyond personal understanding, without a professional post edition.
Consider devices such as Ili or Travis the translator. They demonstrate the usefulness of machine translation in providing a basic understanding. The manufacturers of Ili state that the purpose of this product is to meet the requirements of travelers who need to make themselves understood. Devices like these, which also have voice recognition, will translate your words instantly so you can communicate with the locals. We’re not looking for perfection. We just want to find a place for dinner.
This brings us to linguistic technology, which helps in the streamlining of projects. In this instance, Computer-Assisted Translation (CAT) systems help the professional translator achieve speed and cost reduction. These systems manage translation memory and guide the translator by leveraging previously-translated text. The memory grows with use and is therefore especially useful for content updates. It maintains higher quality than machine translation. It can also be a great help for specialized translations, since the text will always be consistent with the terminology used. Other systems include portals and connectors that monitor the progress of a project.
There’s one more very important aspect to consider, in this post-human world, when it comes to Translation. Although MT can spit out vast amounts of content in all the common languages of the world, it cannot localize that text. The considerations of culture, of different societies and their norms, are simply beyond any AI. So far.
So, language technology continues to evolve, and we continue to translate ever-increasing numbers of Words. We expect not only accuracy, but also understanding. Professional translators are still necessary if we want quality results.
Following this philosophy, at MondragonLingua we will always take advantage of technology to offer our clients the best possible service, but maintain the personality and credibility of a work reviewed and supervised by experienced professionals.