Artificial Intelligence Can’t Keep Up

Creative geniuses in each language come up with unique ways to use words. How to translate the meaning to other languages creates another level of complexity and sometimes artistic license in interpretation. So… good luck with your neural machine translations! [sg_popup id=92]
[us_image image=”24552″ align=”center”]

Speech Isn’t an Exact Science

It is a creative process that can conjure images, feelings, and can cause people to even taste and smell things that are not present in their immediate environment. Likewise, the interpretation or understanding of what is said or written varies greatly, even among people who speak the same language.
[us_image image=”24562″ align=”center”]

Chatting with Siri

For anyone who has started to use their voice to communicate with their phone, car, television or other home automation device, you quickly learn that you have to talk to computers differently than you do humans. I call this “computer speak.” Alexa? Siri? Hello!
[us_image image=”24564″ align=”center”]

Neural Machine Translations Struggle with Trends

New words are coined every year and old words get used in new ways with new meanings. On average, Oxford Dictionaries add approximately 1,000 new entries each year to their English Dictionary.
[us_image image=”24569″ align=”center”]

Yes… Context Matters

In the Smurfs comic, the characters replaced adjectives, adverbs, verbs, and nouns with the word smurf or some form of it in almost every sentence. Yet a 5-year-old watching the cartoon version knows exactly what they said with this frequent word choice. The 5-year-old mentally replaces the “smurf” word with the right word without missing a beat, even if the word was never used in the episode. All the words were defined by the company they kept. Context is everything, which is a challenge for neural machine translations.
[us_image image=”24570″ align=”center”]


Language services industry veteran, entrepreneur, senior executive and visionary, Tim Hunt manages technology, innovation and development at inWhatLanguage for the evolutionary UNIFY translation cloud platform.
[us_image image=”24572″ align=”center”]
To learn more about neural machine translations, download the whitepaper for all the details from Tim.