This is how machines are transforming the translation industry

Towards a world without foreigners: this is how machines are transforming the translation industry

When Google Translate was first introduced, it only supported two languages.

Now, Google Translate supports 133 languages ​​and is used by hundreds of millions of people daily. These advances are made possible by the development of machine translation and artificial intelligence (AI).

How are AI and machine translation used by translation service providers today? What do translators have in store in an increasingly automated world?

State of machine translation and AI today

Will Artificial Superintelligence Create Infinite Economic Growth or More Inequality?

The use of software in the translation of texts is gaining ground. Neural-based engines now mimic the functioning of the human brain, and just like humans, they are designed to learn over time. They do not rely on grammar rules to produce translations. Instead, these engines try to understand the context of what was said and use that understanding to produce a translation.

Neural machine translation (NMT) is much more capable of understanding the intent of the original text. This means it can produce a more accurate translation and read as originally intended. In other words, it comes close to doing translation the same way humans would.

How translation services use machine translation today

Translation using software has progressed rapidly. It has become more accurate and is still being refined.

Yet the translation produced by machines and software still has a long way to go. At least compared to the translation produced by humans. For example, while millions of people use Google Translate every day, there are still many translation projects where it cannot be used.

In an article published on The Verge, it was claimed that Google Translate is unreliable when it comes to medical translation. Sometimes this may work for commonly used languages, but for other languages ​​it may give misleading information.

A 2019 study found that Google Translate was 90% accurate for Spanish. Still, that 10% inaccuracy is too high when you’re working on medical information. Accuracy decreases when dealing with a less commonly spoken language.

The same is true when using software to translate highly specialized text. Like legal documents for example. Relying on a machine might not be the best option as there might be legal intricacies that the software might miss. The consequences of such an error can be serious.

Another pressing problem facing machine translation is that certain expressions can be mistranslated because a machine cannot understand the cultural context in which they are used. This is where localization is so important. The social and cultural framework of the source language must be understood and then transposed to the social and cultural framework of the target language. Computer software can do a lot, but it hasn’t reached the level where it can handle that.

But translation services today use software translation in a different way. The text goes through translation software and the result is then reviewed and corrected by a qualified translator. With this method, you can get the best of both worlds. You can get the speed of software translation and the accuracy of human translation.

Some companies prefer this method when they want something fast but still accurate. The truth is, nothing can beat the accuracy of a translation performed by an experienced translation service provider. They have the language specialists to take care of the right kind of translation needed for the project.

It is a common misconception that language skills are all that is needed for successful translations. In reality, specific technical knowledge may also be required. In the example cited above where Google Translate is still not reliable in translating a medical text, the best solution is to work with a translation service provider who can send specialists who are also linguists. It can also be the other way around as long as the service provider can provide language solutions tailored to the requirements of the project.

Translating with machines and AI in the future


Translation with the use of machines and software will not lag behind human translation. In fact, it catches up quickly. Because the AI ​​translation is designed to build on its knowledge, it can continuously improve.

The biggest advantage of translation with the help of AI is its ability to process large volumes of text in a relatively short time. No human translator can compete with machines when it comes to translation speed.

But human translation is unlikely to be surpassed when it comes to transcreating emotional and thought-provoking texts. Humans will also always be needed to ensure that the context of a text is understood. For now, AI and translation can work together to break down barriers. It will take some time before AI can fully hold its own when it comes to providing translation services.

James G. Williams