Jump to content

The Evolution of AI Translation: The Future of Translation

As part of our series on the future of translation for 2020 and beyond, we're exploring how AI is enhancing human translation like never before.

This January, we’ve been excited to launch our brand new blog series: The Future of Translation. 2020 has marked the beginning of both a new year and a new decade, giving us the perfect opportunity to explore what’s next for the language services industry. Last week, we looked at the top 4 emerging trends for LSPs in 2020. This week, we dive into the evolution of AI translation.

We’re living in a time of great transformation.

Between separatist movements dominating global politics, emerging economies set to continue their rapid growth and the seemingly-unstoppable tide of globalisation, one thing is certain: For everything to stay the same, everything must change.

It's true: The evolution of the internet has led us closer to the total democratisation of information. But at the same time, the decisions of the few now have higher-stakes and a greater impact than ever.

What does this mean? It means that international communication has never played a more important role than it does today.

And in a world of over 7,000 spoken languages, language services providers are fundamental to facilitating this communication across the globe. But what impact will the aforementioned technological advances have on cross-border communication?

A stretch of machine code behind a pair of glasses symbolises the relationship between AI and human translators.

Machine translation and the evolution of AI

One aspect of the translation field that has evolved dramatically in recent years is machine translation. Machine translation, also referred to as MT, automated translation, automatic or instant translation, is simply defined as, “the process of changing text from one language into another language using a computer.” Founding research in the field of machine translation can be traced as far back to 1949.

But... To understand how machine translation works, we first of all must consider the cognitive process behind human-led translation. How exactly do translators work?

When a translator is given a piece of text in its original language (called the ‘source text’), they need to decode the meaning of the text and replicate its meaning in another language (called the ‘target text’). Although this sounds straightforward enough, a professional, experienced translator will approach the task in a very specific way. Armed with an in-depth knowledge of the grammar, semantics, syntax and idioms of both languages, they’ll come into battle ready to face all the idiosyncrasies of the text, including cultural and regional nuances.

Traditionally, machine translation has worked on the basis of simple substitution of words in one language for words in another. However, given that it functions on input, machine translation has long lacked the contextual and idiomatic knowledge to rival human translators.

That is until recently.

An outstretched hand of an AI robot.

Types of Machine Translation

There are three key types of machine translation:

  • Rule-based systems use a combination of language, grammar rules and dictionaries for common words. Specialist dictionaries are created to focus on certain industries or disciplines to ensure that the terminology remains consistent as well as relevant for each industry..”
  • Statistical systems do not have ‘knowledge’ of language rules. Instead, the systems ‘learn’ to translate by analysing large amounts of data for each language pair. As it goes on, the MT begins to learn more about the language pair. Statistical systems can be trained for specific industries or disciplines as well by using additional data relevant to the sector needed.”
  • Neural Machine Translation (NMT) is a new approach [...] that makes machines learn to translate through one large neural network (multiple processing devices modelled on the brain).”

There are also hybrid MT systems that combine systems to get the best possible translation accuracy.

Where does AI come in?

Machine learning (ML), a key concept of AI since its humble beginnings, is the study of computer algorithms that improve automatically through experience.

According to Tom Mitchell,

“A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.”

This means that AI-powered translation is built to continually improve its output. So, what does that mean for the way we communicate? And how does this impact language service providers?

Using AI Translation: The practicalities

AI translation is truly unparalleled as cost and time-saving tool.

Using raw input, AI-powered translation has the ability to build upon its knowledge and train itself to improve its output.

However, language services providers will often then task a human translator with improving the completed translation (called Post-editing) to guarantee accuracy.

But, as AI-powered translation improves and learns, there is less and less need for heavy post-editing. From a user’s point of view, it means that investing time in the AI tools from the outset, such as creating translation memory banks of commonly-used terminology in your sector, directly translates into greater cost and time-savings in the long run.

You still have the option to task a human translator with post-editing but, as the AI learns and becomes more intelligent, the overall post-editing process will become more efficient.

The future of AI translation

Ultimately, then, AI translation can fulfil two main roles: Firstly, as a standalone tool to translate in bulk, as well as a tool to assist human translators, increasing their productivity and daily output and helping them hit tight deadlines.

In the near future, we are convinced that AI translation will allow clients and linguists to interact more efficiently, ensuring we can make the most of the finite time and resources of a skilled human translator.

Signs of speed limits symbolise the limitations of AI

AI translation: Its limitations

It's no secret that every new technology has its limitations.

AI-powered translation is ideal for large volumes of content that are required for understanding purposes only. However, it is unlikely that it could ever rival a human translator in translating and transcreating emotive, impactful and persuasive copy – The kind of copy that's needed in marketing, advertising or high-stakes communications.

Creativity is still very much the domain of humans, and that’s unlikely to change anytime soon.

But one thing's for certain; AI-powered translation isn’t going anywhere – and it will continue to grow at a rapid rate. That's not to say that Google Translate is suddenly about to be adopted by global businesses across the world – AI translation is set to be rooted instead in bespoke technology, specifically tailored to the client’s needs, fostering an on-going collaboration between AI and human translators.

Why Wolfestone?

Here at Wolfestone, we’re experts in providing high-quality translation services and we have a track record of helping our clients successfully go global.

Get in touch with our expert team and find out how we can help you reach your goals.

Get a quote
CTA

Get the latest translation insights straight to your inbox