New heights for Google, Facebook & Microsoft’s language translation
Artificial Intelligence has long been applied to enhancing human communication. Interacting with other people is an integral part of life, but one that can be complicated when messages are quite literally lost in translation. Various services have been developed to address global language barriers. Last year, Google announced that they would replace their old service with one that demonstrated a much higher success rate. Since then, the company has already released the new system for Chinese to English translations. As always, other tech giants are hot on their heels. It’s claimed that Facebook’s new system is nine times faster than any other translation service, and Microsoft is directly targeting entrepreneurs with a new PowerPoint translator. How might this change the way that we communicate, and how will these new systems impact different industries?
Human level translation. . . almost
Just how accurate is Google’s new service? In a report published in September 2016, bilingual people were asked to compare Google translations to human translations. In English to Spanish translations, Google scored 5.43 out of a possible six, whilst humans scored 5.55. Yes, the humans were more accurate, but the software was really not far behind. This is a huge improvement on the previous service, which hardly came close to the human translations. In fact, the new system scored 64 to 87 per cent higher than its predecessor. Competitors like Facebook have made real improvements too. As only half of users actually speak English, Facebook’s ‘multilingual composer’ aims to eventually enable conversation between anyone in the world. As of 2016, Skype’s translator now works for calls to landlines and mobiles. Last week, Microsoft demonstrated PowerPoint Presentation Translator, a machine learning tool for business presentations that auto-translates slides into languages including Hungarian, Czech, French, Chinese, German and Spanish. It also translates the presenter’s words on screen in real time. Unlike previous methods, these systems use machine learning to improve translations with context. Somewhat worryingly, though, it’s not entirely certain as to how the software makes its decisions. Google researcher Quoc Le has admitted that black box systems can be unsettling, but added that “it just works.”
How will accurate translation affect the way we communicate?
The availability of high quality translation will be incredibly positive in terms of human communication, as it will contribute to breaking down the language barriers which can complicate verbal interactions. This will, in theory, make it easier to be understood. In an important, international business negotiation, making sure all parties understand each other is absolutely vital. There is an obvious consumer level here, too. It will become easier for people to travel, making sure they can be understood by native speakers. . . until they lose their phone, at least. In many ways, accessible and accurate translation will be helpful to corporations and individuals alike. However, taking away the need for translation takes away the need for a translator, which will have a detrimental impact on employment. Why bother with language teachers in schools, for example? International companies and global organisations won’t have to use professional translators, and on a national level, tour companies won’t need to hire a French speaking guide to deliver French tours. Tourism, particularly in economically struggling countries, provides employment opportunities which could be stripped away by the removal of language barriers. It could even be argued that on demand translation will make humans worse at communicating, much like smartphones which enable interaction but are often regarded as antisocial.
On demand translation at a human level will bring peace of mind to holiday makers, security to businesses and clarity to any other multi language exchange. But not all implications will be positive, and developing software to the point that it can match human ability is going to have obvious consequences for employment. Even so, Google Translate (and similar services) won’t remove the importance of learning other languages. Especially in business negotiations, using just a few words of someone else’s native tongue can be endearing. In some cases, it may even be offensive to whip out your smartphone instead of attempting to speak a language yourself. As accurate as new services may be, the ability to speak other languages will always be a coveted human skill.
Could you improve your business through accurate, real time translation? Will human level translation make it less necessary to learn other languages? Which other professions could be disrupted by translation software? Comment below with your thoughts and experiences.