Chatbot and the Language Industry

By: Cheryl

 What comes to mind when you see the word chatbot? Siri? Alexa? Or the automatic replies you receive when you type in an online customer service chat box? Well, they are all a part of the big chatbot family.

As the name indicates, chatbots are robots that can automatically generate conversations with humans and fall into the artificial intelligence (AI) category. When asked questions, chatbots generally come up with answers spontaneously, which is a typical way of interacting with humans.

As social creatures who, by instinct, seek interactions and communications, humans’ quests for robots that can “talk” to us date back to the last century. The first chatbot that ever made its presence in history was ELIZA, developed in the 1960s by computer scientist Joseph Weizenbaum in the U.S. With limited knowledge reservoir and capacity, ELIZA made conversations only by taking sentences from the user’s input and turning it into an interrogative form. Simulating a psychotherapist’s operation, when being told “I’m unhappy,” an expected response from ELIZA would be, “What made you unhappy?”

Following ELIZA’s path, several different chatbots have since emerged, such as PARRY, which tried to develop a “personality” and come up with emotional responses; Jabberwacky, which simulated natural human chat with a sense of humour; and ALICE, which is open-sourced and uses natural language processing, allowing for more sophisticated conversations. The revolutionary step for chatbots, however, was the development of SmarterChild in 2001, when for the first time, a chatbot could retrieve information from databases and help users with practical tasks such as the weather, news and stock prices. Humans’ creativity in chatbot development has since then taken flight. To name a few chatbots that we’re familiar with today, Siri emerged in 2010, Google Now was invented in 2012, and Alexa was developed in 2014.

There are different ways humans can interact with chatbots. Some chatbots are button-activated, and users click on the built-in buttons or menus to trigger a response; some require users to type in the keywords by themselves, and some are more “intelligent” in that they can extract the information they need in order to respond to natural human language, either by typing or by voice.

A chat with Kuki.

The primary goal of a chatbot is usually to assist humans with tasks or information, although there are also chatbots designed for other purposes. Alongside the usual chatbots that can help us set an alarm, book a table at a restaurant, search with keywords or act as a handy customer service representative that can respond at any time, there are also chatbots that seem to have been invented just to kill time or for fun. Ometrics Joke Bot, for example, is designed to generate random jokes when a button is pressed. Kuki, an award-winning “superintelligent robot” in her own words, can detect natural language and chat with you on whatever topic you’d like.

Kuki speaks English, but understands some fundamental words in other languages.

Not only does Kuki understand some basic words in different languages, she is also able to tell when there is more than one language used in the same sentence. In fact, this is quite remarkable, as the multilingual feature for chatbots is hard to achieve, especially for a context-based model. In today’s world, where local people tend to mix different languages in their daily communications (India, for example, where over 100 languages are spoken in the same country), detecting the languages separately and coming up with proper responses still project their challenges. This is where a heavy load of localization work must weigh in.

Kuki detecting two different languages in one sentence.

For chatbots that are truly multilingual, there comes a rise in popularity. They are now commonly used in the field of customer service, and there are countless articles on why businesses should consider using them or guidelines on how to build one of your own. Another way that people find chatbots extremely useful is in language learning. These robots can simulate conversations and talk in natural human languages with impeccable spelling, grammar or pronunciation, making them the perfect companions in language learners’ journey.

From button-activated to context-based, task-based to conversational, monolingual to multilingual, chatbots have come a long way, and their bond to the language industry has been ever so close. How will chatbots keep evolving in the future, and where does this evolution take us? When technology meets language and culture, truly anything is possible.