Thanks to deep neural networks, global, seamless communication isn’t far off.
In 196 BC, the Decree of Memphis was issued by Ptolemy V in Egypt. It was written on a granodiorite stele (a big, flat, upright stone) in three scripts. The first in Ancient Egyptian hieroglyphs, then in a Demotic script and finally in Ancient Greek. It proved to be the key to understanding the hieroglyphs thanks to its three identical texts; we know it as the Rosetta Stone.
Thanks to Microsoft, we’re on our way to a universal translator that will make communicating in several (and, presumably, eventually all) languages seamless. Effectively, Skype Translator is the Rosetta Stone of the technological era.
I’ve written quite a bit about my amazement at the world of machine and deep learning. This is an example of deep learning in action. Conceptually, a universal translator has been around for decades (long before modern computers came into existence).
The first patents filed for this kind of technology date to the 1930s. And while so many have tried, most have resulted in failure, while the best saw minor successes (if any).
Now, Microsoft (Skype) is aiming to be the first to perfect the concept that had once been a thing of science fiction.
How It Works
I won’t pretend to be a brilliant engineer behind the technology. But I can explain how, in the most simplistic of terms, this works.
For those of you unfamiliar with the concept of machine learning, it is more or less exactly what it sounds like. The complex systems and algorithms that make up an application are somewhat intuitive, adapting to nuances, frequencies and, in the case of natural language processing, things like colloquialisms, to become more accurate. Essentially, the machine evolves as it is exposed to more and more data, helping it become sharper in its results.
In an interview with Time magazine, Lane Schwartz, a linguistic professor at the University of Illinois explained, “The more data you have, the better you’re going to do.” It’s that simple.
What Microsoft has done (according to its engineers and product teams) is perfected the first form of the universal translator that will use what it learns in beta testing to become smarter, more intuitive and faster.
Ultimately, when one person speaks in their preferred language to another person with a different language setting, the system will (almost instantaneously) register what was said and (with grammatical accuracy) regurgitate the phrase in the other user’s language.
Pretty impressive stuff.
If this works as well as demonstrations have shown and Microsoft has promised, it will (quite literally) unite the world one front.
We often hear the jokes about fairly trivial apps ‘changing the world’. Well, this actually would. On the business front, this would be a first step in the convergence of several currently detached economies. I say first step because, while communication and those items lost in translation matter, there are still major social hurdles to overcome (in the way different societies do business).
The ability to provide aid and assistance from a distance or during an emergency is also greatly facilitated with this kind of an application. Imagine someone calls you frantic in a language you don’t understand, but your universal translator is capable of translating it so that you can understand it instantaneously.
And those are just a few of the amazing things that can be done.
It’s exciting to think about how this will make the world smaller (yet again).