AI vs. Translation: Let’s get back at it
Language is a barrier to growing businesses. Duh, what a no-brainer, right?
Enter translation, the grunt in charge of the heavy lifting for companies all around the world. In a globalized world, translation businesses thrive and offer a variety of highly diversified services. But even translation businesses cannot deal with all problems, which is where localization, our silent hero steps in. Imagine, the challenges certain apps create when having to be launched in different markets. Fine-tuning and a deep knowledge of the country in cause are essential. Yet not all companies have the necessary budgets or the will to invest in businesses which offer integral localization solutions.
And yet, the global markets are waiting to be reached, but the technology behind the services in question is just not there yet. AI is beating around the bush a little and has some difficulties processing string requests in one language. Even Google has been at odds with its neural network translation service, which as potent as it might be, still makes some funny translations.
Michael Housman, Chief technology officer for RapportBoost.Ai has been debating the ideal scenario for ML and AI. He’s figured that AI works best with a set of fixed rules, naming chess and Alpha Go as go-to examples. A machine beating a human at GO, was a full-on success, precisely because the algorithm behind it, thoroughly knew the rules behind the game.
“Language is almost the opposite of that. There aren’t as clearly-cut and defined rules. The conversation can go in an infinite number of different directs. And then, of course, you need labeled data. You need to tell the machine to do it right or wrong.(…)Two translators won’t even agree on whether it was translated properly or not.(…)Language is kind of the wild west, in terms of data.” — Michael Housman
Housmans approach with his company RapportBoost.AI is to create an emotional intelligence engine, which guides companies through communication nuances, rather than focusing on strict sentences. Therefore, he asks:
“ What are the attributes of the conversation that contribute to a good outcome? It’s looking at hundreds of different variables. Right now we are solely focused on human agents.`Cause you know the chabot market isn’t quite there yet.But we make a habit of testing out every bot we can. And, you know, kicking the tires. And just generally, it’s not there. The technology is not super sophisticated. It can’t come up with answers on the fly. Oftentimes, just responding in the appropriate manner is a challenge.”
With a sigh, he dismisses chatbots for failing to relate to people on an emotional level: “The EQ seems to go under the radar. People don’t think about it as much but we’re naturally social beings and that stuff matters a ton.”
Having an eye on future developments in the language and chatbot field , Mister Housman, concludes: “I think that we’re going to see a massive uptick in the extent to which bots are used for sales-facing endeavours. Right now, most bots have been programmed to engage in customer service. That’s going to change quickly and we’re going to see a lot more instances of Alexa-like platforms where you can tell a bot what you’re looking for and it’ll offer up useful recommendations and, ultimately, complete the transaction. I read somewhere that conversational commerce is expected to grow to a $600 billion market, which sounds insane but when you consider the growth of virtual assistants, bots and voice-activated bots like Alexa, I actually think it’s pretty reasonable.”
Houseman also hands down reasonable advice, when working with AI. He believes, that as cool AI might be, it shouldn’t yet become the sole driving force or focus of companies, but rather more a tool to ensure performance and profitability. Having a “human in the loop”is a good start, yet we will have to diversify the field of human-robot interaction, so that we, won’t end up performing menial tasks.
Sentencing and such
In another space-time-continuum Erik Cambria, academic AI researcher and assistant prof at Nanyang Tech in Singapore focuses on NLP, the base of AI language translation. He analyses the reading comprehension of AI and believes that there are a lot of things humans unconsciously do while absorbing a text:
“The biggest issue with machine translation today is that we tend to go from the syntactic form of a sentence in the input language to the syntactic form of that sentence in the target language. That’s not what we humans do. We first decode the meaning of the sentence in the input language and then we encode that meaning into the target language.”
Dr. Jorge Majfud, Associate Professor of Spanish, Latin American Literature, and International Studies at Jacksonville University, also wanted to explain why AI is so buggy when it comes to language. His conclusion was that, a machine glitches out due to a faulty mindset. It considers an entire sentence it has to translate. Meaning can thus be transported in relation to a paragraph, the rest of a text, a specific word, a cultural context or even its speakers intentions. Certain language variations or tropes evade textual translations.
“Google translation is a good tool if you use it as a tool, that is, not to substitute human learning or understanding.”
Finally, when playing around with AI in localisation and translation, we have to look at some other factors. Such as cultural risks, one takes when inputting a certain text into a machine. We’ve been talking about how AI transports biases or how machines can be programmed to reflect the beliefs of their creators. Dr. Ramesh Srinivasan, Director of UCLA’s Digital Cultures Lab, is a strong believer that said translation tools should be transparent about their capabilities and limitations:
“There tend to be two parameters that shape how we design ‘intelligent systems.’ One is the values and you might say biases of those that create the systems. And the second is the world if you will that they learn from. You know, the idea that a single system can take languages that I believe are very diverse semantically and syntactically from one another and claim to unite them or universalize them, or essentially make them sort of a singular entity, it’s a misnomer, right?”
So, to sum up, the way for AI to master translation is to get rid of the programming bias, get some empathy crash-course and learn the rules, that there are no rules in language! :D
If you like our articles, make sure to let us know. If you don’t, well then, make sure to let us know too.
About Beluga
Beluga helps fast-moving companies to translate their digital contents. With more than a decade of experience, professional linguists in all major markets and the latest translation technology at use, Beluga is a stable partner of many of the most thriving enterprises in the technology sector. The business goal: To help fast-growing companies offer their international audiences an excellent and engaging user experience.