Don't Press One For English: AI & Multilingual Support

There are over 7,000 languages spoken on earth and no human who can speak or comprehend even one percent of them. In fact, the world’s most pre-eminent polyglot, Ziad Fazah speaks 58 languages — an impressive feat but nowhere near comprehensive enough to communicate with a global population. AI, on the other hand, has vast potential for learning every language on the planet — even those without large corpora (structured sets of tests) available, thanks to a recent natural language processing (NLP) breakthrough from Google.

The tech behemoth recently detailed NLP advances in a preprint paper accepted to the Interspeech 2019 conference in Austria. By applying AI-acquired knowledge from data-rich languages (such as English, Arabic, Hindi, etc.) to data-scarce languages (dialects or languages spoken in isolated communities) the company was able to build a multilingual speech parser that can recognize nine different Indian languages.

The use cases for such a system are profound: in countries like India, with a population that speaks over 30 languages, each of which is well represented, a single multilingual model that could understand each of these languages could be applied across a wide range of industries, and could serve to democratize access to information around the world.

Google is likely to leverage this system for its Google Assistant, which has multi-turn multilingual support for English, Spanish, French, Korean, Hindi, Danish, Dutch, and Swedish, as of February, as well as Interpreter Mode for dozens more languages. As NLP and natural language processing advance, Google won’t be the only one leveraging technology to provide multilingual support.

The Future of Multilingual, Global Support

Many companies have already integrated forms of AI, such as bots and automated ticket routing into their customer support processes. However, when global companies don’t have access to multilingual automation, they are faced with a logistical and financial barrier to offering superior support. Staffing around the globe for every single language that a customer might speak can be enormously costly, especially when the volume of tickets in a given language is low but customers still expect round-the-clock support. However, the alternative of offering only English support can be a barrier to growth — which leaves many companies between a rock and a hard place when it comes to multilingual support.

This is precisely the issue that advances in NLP will solve. Bots can currently handle 60 percent of all incoming tickets without any agent assistance, which provides enormous cost-savings to companies and efficiency benefits to customers. The only people who realize these benefits, though, are those who speak the languages that the bot can comprehend.

Automated language comprehension will usher in a new era of support, one where bot-based messaging and smart speakers become primary contact modes for customers around the globe.

Are We There Yet? How to Prepare For Global CX

As with any impactful emerging technology, those who get on board early will gain the biggest competitive advantage. To prepare for automated multilingual support, brands today should be focused on two key infrastructures:

1. Messaging Based Support

Voice-based support is an incredibly new medium that hasn’t yet been fully explored, but messaging is a fully functional, up-and-running channel that consumers already love. In order to fully reap the benefits of multilingual support, brands need to already have a strong messaging foundation.

2. Bots and AI

The more familiar a company and its customers are with bots and AI, the easier the transition will be down the line as these two technologies gain importance in the industry. Many of the largest companies in the world, including Amazon (which is also working extensively on advancing NLP), leverage bots and AI to enable customer service to scale with both a growing user base and a rapidly evolving technology landscape.

The pace of technological advances, particularly when it comes to AI and automation, has never been faster. The companies that come out on top of this period of rapid change will be the ones that don’t leave their customers behind in an outdated, analog world.