AI and localization: Transforming, not replacing, human expertise

Unless you’ve spent the last two years on a desert island, you can’t have missed all the discussion about AI taking over multiple tasks across various businesses. This discussion has been particularly interesting in the localization industry: will human translation become irrelevant?

 

The short answer is no.

 

Although all the hype around AI is beginning to slow down (as companies realize the value of humans in the loop), new tools and solutions are here to stay and have already significantly changed the way we think about automation and creativity.

 

There are visible benefits of incorporating AI solutions to automate repeatable tasks and tasks that require pattern recognition.

 

I recently got round to reading ‘Thinking, Fast and Slow’ by the Nobelist Daniel Kahneman. In one of the chapters he describes research conducted on a group of parole judges. The study discovered a statistically significant difference in paroles that were granted after officers had enjoyed a food break, like a second breakfast or lunch. Even more interesting is that the number of paroles granted declined just before those breaks. This example illustrates how our environment and current state of mind affects the decisions we make.

 

I couldn’t help making the connection in my head about how AI could be used to support decisions, making us less dependent on environmental influences – while at the same time allowing humans to concentrate on more nuanced cases and use judgement where necessary.

 

This brings us to the localization industry and role of translators in the process. For some languages and content types Large Language Models (LLMs) are capable of increasing the quality of initial machine translation output in comparison to pure neural machine translation output – which has so far been the technology standard used in the localization industry.

 

LLMs are being used to verify the quality of machine translated segments and judge if it is good enough to meet client requirements. This means that translators could ultimately translate less content directly. Does that mean that businesses will need fewer and fewer translators?

 

No.

 

It turns out that similarly to the example with parole judges, translators could benefit from a LLM to handle simple translations, allowing them to concentrate on more complex or nuanced content.

 

AI solutions change the scope of work for translators, giving them more work that requires nuanced, subject matter expertise, and involves more research and background. It makes the work of translators more valuable than ever. Contrary to what might be an easy conclusion that translation has now become completely commoditized, translation services are evolving towards linguistic consulting and translators are becoming Language Specialists, empowered with more and more technology.

 

Given the content explosion businesses will not need fewer translators, but they will see a shift in their role and expertise. Restating my thought from the beginning, businesses are realizing the importance of humans in the loop when applying AI solutions - and the same value is applicable to the localization industry.