Practical Uses for AI Beyond Machine Translation
Share
Click here to close
Click here to close
Subscribe here

Practical Uses for AI Beyond Machine Translation

How artificial intelligence is shaping the localization industry is a subject we approach from many different angles.

Generally, though, we’ve found that those interested in AI tend to fall into three camps: those who are suspicious and fearful of it (Will AI replace humans?), those who directly develop the technology (How can we improve AI?) and those who are optimistic about its potential even if they don’t know what that looks like (How and when should I deploy AI, and what for?).

If you’re in camp three, we have some insight we think you’ll find useful. Jim Compton, Manager of the Partnerships Program at RWS Moravia, recently wrote an article for MultiLingual Magazine about the “killer applications” of AI that can solve practical problems in several areas of localization—not just machine translation. Below is a summary of his ideas (and your free copy of the full article).

AI: The story so far

First, some context: what has AI already helped us achieve?

AI has always been part of the localization culture. Beyond machine translation (MT)—which we’ve been using since the 1960s—we’ve been using Translation Memory, concordance searching, terminology recognition, optical character recognition and other applications of AI for some time.

The question now is how to apply AI to even more localization challenges, which of course have evolved just as quickly as AI itself. Our goal of crushing language barriers hasn’t changed. But the continued shrinking of the world and the rise of digital communications and other global megatrends have made it harder to do—and many agree we need AI to get there.

Jim likens the use of AI in loc to having extra hands on deck. “What would you do with 150 extra engineers?” he asks, or in other words: how would you solve these challenges if you didn’t have human resources limitations, whether they be in translators, project managers, copywriters, solution architects or account managers?

Rethinking how to apply AI

It’s best to approach this question by looking at specific situations you’ve found difficult to solve. When have you been forced to waste resources? When did you have to compromise on a solution because of human-based limitations?

One of these scenarios might sound familiar:

  1. You archive valuable transaction data after completing a project, only to have to re-learn the same lessons in later, similar projects.
  2. You estimate the price of the project with a “finger in the wind” approach—and sometimes find it to be inconsistent with a similar, previous project.
  3. After developing a solution to what appears to be a “unique” client challenge, you find out that another team already developed another (better) solution to a similar challenge months ago.

What do these situations have in common? They could all benefit from applied intelligence. Whether it’s to better recognize patterns, understand cause and effect or predict future problems, throwing humans at the issue would be too expensive and might not yield the results you desire.

But what about when you can’t even identify the problem? How can AI help then?

Think of it like Twenty Questions, Jim suggests. You know the game: write something on a sticky note, stick it to someone’s forehead and that person has to guess what’s on the note by asking a series of quantifiable questions. Port this concept over to localization and you can ask questions like, “Have we seen this situation before?” and, “If so, what features, characteristics or factors define it?” to help narrow down possibilities, differentiate one problem from another and identify existing best practices or gaps that AI can fill.

Putting AI into action

This application, or what Jim calls “opportunity profiling,” is just one way he’d make use of 150 extra engineers to boost productivity and resolve business problems. Still, it’s not easy to navigate the diverse landscape of AI technologies—SaaS systems, API-based cognitive microservices, open-source toolkits—to find the best solution for you.

So, Jim went on to share with us the process (including the code) he used to set up a working example of opportunity profiling and the solution he envisioned. It’s a method you can apply to your own thinking about how to better optimize your localization program. He also includes future AI trends and industry leaders’ input on practical uses of AI in loc.

Download the article and let us know what you think!

 

Comments