Robots don't speak Zulu. Will AI be the death of our indigenous languages?

Smart assistants Siri, Bixby and co are lovely. With the push of a button you can ask them stuff and a cheery robot voice will trawl the internet and almost instantaneously come back with an answer. It's like having Hermione Granger's brain available to you at all times. There is just one problem: your robo-Granger knows everything except the words coming out of your mouth. 

We've all had that frustrating encounter with our digital assistant where you fire her up, ask her a question and watch as she gives you the answer to something completely different. If you ask where the nearest coffee spot is in heavily accented English, she (or he, depending on your settings) is as likely to send you to a Starbucks as she is to redirect you to a racist Wikipedia page.

If you're not some type of European then you can more or less forget trying to communicate with your smart assistant in your mother tongue. The idea of Siri speaking Venda seems as close to being realised as Elon Musk's hyperloop. This brings up an interesting question: for all artificial intelligence's (AI's) wonderfulness, could it inadvertently be contributing to the death of languages?

According to a report published by the Catalogue of Endangered Languages, of the roughly 7,000 languages spoken in the world today, 46% are endangered, and about 335 of those have fewer than 10 speakers.

More than 639 languages are known to have become extinct and 227 of those deaths have occurred since 1960.

Distressingly, that death toll increases by about one every three months. All of this is very sad and deserves an orchestra of tiny violins, but what does it have to do with AI?

We live in the golden age of convenience. Depending on what part of the world you are in, a simple word in Alexa's ear and within a few hours some chap in an Amazon T-shirt will knock on your door with the groceries you asked for.

Alexa is not the only one who can weave this type of magic. Perched merrily above a factory near the coastal city of Busan in South Korea is a vision of what a truly smart home can be. It is the Jetsons brought to life. Powered by a harem of LG products, the home features a fridge that can show you what's inside it without you having to open the door, televisions that disguise themselves as artworks, washing machines that can be activated remotely and vacuum cleaners that do the job automatically.

Managing all of this is CLOi, an LG smart assistant who operates similarly to Alexa. If you have everything synced up just right, good old CLOi can check in with your fridge and find out if you need milk, fire up the washing machine so that your delicates are clean by the time you get home, or Shazam a song playing on the TV - provided you say the right words - and finding them is trickier than it sounds.

"Market share and difficulty is what determines what languages we make available," said Ken Hong, LG's senior director of global corporate communication. "Ease of translation is also very important. Mandarin, for example, is very difficult in translation. That is why you won't see Mandarin in many AI plans and strategies, in the beginning. It takes time."

What also complicates matters is how AI, and by extension smart assistants, learn. In a nutshell, CLOi, Siri and co learn languages and accents by having a massive data set of that language from which they can pick up patterns. That means that, much like humans, the more they are exposed to a language, the better their grasp of it will be.

Unfortunately, there just are not enough people speaking Zulu to Bixby for her to bother learning it. Companies are not incentivised to create products that communicate with people in their native tongues if those people do not clamour for and then buy them.

One of the unintended consequences of the convenience of smart homes and devices is that a certain level of prestige is granted to only a small set of languages. If you are a native Setswana speaker but have to use English to do your online shopping, then the message is that English is the more practical language. Not only that, a certain type of English will get you further in the world than the language you grew up speaking.

Many of us will be familiar with the received wisdom at school that it was better to take up something like French rather than an indigenous South African language because French travels better.

None of that, however, is the fault of AI. For the time being at least, artificial intelligence is just a tool. One to be put to use as we see fit. In certain parts of the world the technology is actually used to save dying languages. Earlier this year, SBS News in Australia reported how the ARC Centre of Excellence for the Dynamics of Language was working with Google to help preserve some of the country's indigenous languages by digitising them, thereby making transcription much easier.

"What artificial intelligence provides is the learning step from audio to the test," the centre's professor Janet Wiles told SBS News. "It sounds like a really mundane thing but that's a key step for language to cross the digital divide, to go from a spoken language to a language where people can text on their phone and use it with digital tools."

As with just about everything technology-related, trying to lay the extinction of languages at the feet of machines is little more than fear-mongering. The machines do what we ask them to do and will continue to do so until the day they rise up and murder us all. Until then, it is important that we ask them to speak to us in our mother tongues.

The preservation of languages is not some kind of ethereal social responsibility project that warms hearts but has little practical value. Different languages express ideas that do not exist in other languages, and having access to those ideas gives the speaker a glimpse into a different world view that has practical applications.

The Japanese idea of "kaizen", which loosely translates as "continuous improvement", has embedded itself in corporate cultures across the globe despite the struggle to find a direct translation for it in other languages.

It is no coincidence that people who are multilingual are considered to be particularly intelligent, or that a number of studies show that bilingualism is good for our brains. Most importantly, however, if we can get the CLOis of the world speaking more than just a handful of languages, we will be helping them do the thing they were designed to do: make OUR lives more convenient.

• Yolisa Mkele visited South Korea as a guest of LG.