Blog

Automatic translation: Why brands need to stop waiting for AI to solve their translation problems

Published on March 19th, 2018

There’s a lot of hype around artificial intelligence (AI) technology right now and there’s genuine reason to be excited. The prospect of driverless cars and predictive diagnosis spotting illnesses at the earliest stages are major technology breakthroughs. Suddenly, it seems like AI will solve all of our technological challenges sooner or later – it’s only a question of time.

However, as with most technology trends during their time in the spotlight, much of the hype surrounding artificial intelligence is overly optimistic and brands hoping for AI to solve all their automatic translation problems might be disappointed.

The AI translation hype

It’s easy to understand why expectations are high for automatic translation and related technologies. The likes of Google and Microsoft have made huge progress in recent years, thanks to breakthroughs in artificial intelligence, and today’s crop of automatic translation tools are significantly better than anything we’ve seen before.

If this kind of progress can continue, why shouldn’t we see translation tools that match human quality in the near future?

The problem is, artificial intelligence as it exists today is nowhere near capable of cracking the language code – and it may never be. Tools like Google Translate use neural network processors to handle vast amounts of data, essentially cross-referencing existing translations to help it handle new ones – and the results are certainly an improvement. However, this technology will always be limited by the use of static data and this makes a number of automatic translation hurdles almost impossible to overcome:

  • Understanding context
  • Language nuances
  • Grammatical accuracy, exceptions, etc.
  • Dealing with creative content
  • Dealing with content that can’t be translated directly

While Google has significantly improved its ability to understand context over the years, there’s only so much it can do. Understanding the context of individual words remains a challenge for the search provider (and other tech giants using artificial intelligence), but things become even more complex when you try to understand the context of full sentences, paragraphs or entire pieces of content.

Even if AI ever becomes advanced enough to interpret contextual meaning on a regular basis, the technology will then have to tackle language nuances like sarcasm, irony and satire. These are all challenges AI needs to overcome before it can accurately understand content, let alone translate it into another language. At this stage, AI technology is far from proving it will ever be able to achieve this kind of sophistication, but automatic translation is still a valuable tool for translators and language pros.

Using automatic translation to improve quality

The idea that machine translation needs to replicate human ability is an unnecessary distraction. Certainly, in the case of artificial intelligence, we’re talking about a technology that needs fixed rules and rigid structures to operate. In these cases, artificial intelligence can often outperform humans unreservedly, forget about matching them.

Languages don’t come with these fixed rules, though, and ten different translators could debate the best translation of a single piece of content. Likewise, linguists could debate the best definition of a single word, knowing there is no “correct” answer to such questions.

All of this aside, if machine translation achieves 25% reliable accuracy, that’s a quarter of the workload handled on behalf of human translators. It’s always faster to quality check and edit translations that start from scratch and automatic translation is widely used by language pros to speed up the translation process and improve quality. This means faster turnaround times and lower costs for the brands investing in professional automatic translation.

So artificial intelligence isn’t going to solve all of your translation problems but it will continue to improve the results of quality translation services.

(Visited 151 times, 1 visits today)
Posted on: March 19th, 2018