Artificial Intelligence

Artificial intelligence (AI) is the ability of a machine to learn and adapt, as opposed to simply processing data based on a set of instructions. In translation and localization, AI has been a kind of holy grail quest, the ability to have a device instantaneously translate language from one language to another.

From machine translation to machine learning

Early efforts at developing true machine translation date to the beginning of the computer era, circa 1955, and have continued since then. But the enormity of the challenge combined with the lack of processing power and memory hampered early efforts, which only resulted in extremely primitive results. However, in the mid-2000s, there was a breakthrough, and it was one of the earliest examples of how artificial intelligence would change our lives.

AI meets neurology, and automated localization makes a leap

Google was one of the first companies to make machine translation available for general use. However, the technology was very limited due to the underlying method used to ‘teach’ the machines the nuances of language including spelling, syntax, grammar, etc. This was rules-based, with the software being programmed to have a set of rules and glossaries to guide it in choosing how to interpret the language. This was extremely laborious, with each language pair requiring the input of large amounts of data and customized algorithms. Progress was slow and results clumsy, at best.

Google and others made significant investments of time and money to reach that basic point. But in Google a small team was taking a different approach, attempting to mimic the human brain’s neurological learning abilities to ‘teach’ AI software. Instead of rote learning based on rules, this gave the machines the ability to learn from mistakes and feedback based on corrections made by humans. The result was a quantum leap in capabilities and accuracy, a leap so large that Google abandoned its earlier investment and went all-in on this new approach.

Multiple languages, increased usability, and a faster view into global markets

The results were a faster and far more flexible translation capability, one that improved as more people used it and it learned from corrections. The software was truly learning and adapting, as our overall definition defines artificial intelligence. It was also able to apply the same rules to multiple languages, simultaneously with much less data input required. The result has been the emergence of a primitive but improving ‘universal translator’.

The challenges of both translation and localization are significant. Context, jargon, colloquialisms, and other factors are challenging when viewed from a learning context. As a result, the results of machine translation and AI are still limited but now improve on their own, a major change in the development of software capabilities. And the implication for translation and other fields is wide-ranging.

What are the business benefits of artificial intelligence?

Translation based on artificial intelligence gives a fast – if basic – window into other languages and cultures. It has become useful enough to offer a base understanding of meaning in content, a base that still requires interpretation by humans to be completely useful. Nevertheless, this base can greatly speed up the translation process, saving time, money, and building a faster ability to enter new markets.

AI already offers many benefits across a variety of businesses and disciplines. It is revolutionizing the reading of complex visual data like images from X-rays and MRIs in healthcare. It is used by the legal profession to analyze contracts and other agreements. And in global businesses, it can accelerate a company’s ability to research and enter new markets.

How to best integrate AI into translation and localization workflows?

Machine translation can be an intrinsic part of the translation workflow, helping project managers and translators understand the basic meaning of content early in the process. The time savings can be significant as there is less parsing of content to understand its usefulness and context, time-consuming processes that have been essential to localizing content accurately.

But it is the view into new markets and cultures that instantaneous translation offers that has the longer-ranging effect. The ability to quickly get a basic understanding of trends, technical content, and market research, among other things, is easier to do by using AI-based translation to give us a rough but more useful context and meaning early on. It enables a faster understanding of the broad global markets.