Automated eLearning translation: Here’s why human-AI collaboration matters (and how Gomo can make it happen!)

What does the much-discussed rise of generative AI tools mean for eLearning content authoring? As with most fields, there are plenty of promising applications, from smoother course reviews to instant content summaries—but we’re especially excited by what this technology means for course translations.

Thanks to generative AI, naturalistic and nuanced automated translation has become more attainable than ever. That’s a big deal. While auto-translation has been around for a while, newer and more sophisticated iterations have followed in the wake of generative AI—and that’s why we’ve introduced a new translation feature powered by OpenAI, the organization behind ChatGPT. Read on to discover:

• How AI translation compares to traditional machine translation
• What to look out for in terms of data security
• The important role human translation and oversight continue to play, and why you need an authoring tool that facilitates it

How AI translation compares to older machine translation tools

Automated translation isn’t a new concept. In fact, machine translation has been an evolving discipline for decades: we’re all familiar with Google Translate, for example. So, what’s the difference between machine translation and AI translation?

In truth, there’s not a clear-cut distinction between the two concepts, and in some ways, AI translation can be seen as a further iteration of machine translation. But, in very broad terms, some forms of traditional machine translation have tended to veer toward the literal. They’ve historically operated based on fairly rigid rules, and, as a result, you might encounter machine translations that opt for a word-by-word approach. That was the conclusion of a small blind study carried out in the UK, which found that Google Translate tended to be quite literal and missed key pieces of context when compared to its AI counterparts.

Let’s look at a common example of this dynamic in action. If you tell a traditional machine translation tool to translate the phrase “break a leg” from English to French, it might choose to translate each word in turn: “break”, “a,” and “leg,” resulting in “casser une jambe.”

If you were to ask an AI translation tool to translate the same phrase, you might get a different result. Today’s powerful AI tools can have a much better grasp of the meaning behind your content, which means they’re better equipped to tackle these idiomatic expressions in a way that makes sense to your learners. In this case, the AI tool might be able to figure out that “break a leg” actually means “good luck,” and offer an appropriate translation like “bonne chance.” This is by far the better choice—especially since the literal translation of “break a leg” might look like an oddly-placed threat to your French learners!

That said, some machine translation models can produce similar results to AI translation tools. But the fact that AI translation is already able to operate to a similar (if not outright better!) standard, and that it’s continuously evolving, marks it out as a highly sophisticated method of translation. And it’s likely to keep getting better.

Fun fact: Gomo’s AI translation can be used in conjunction with different language layers

AI translation is characterized by efficiency. It’s all about quickly securing a naturalistic and sophisticated translation without needing to involve manual processes or human translators. But here’s the thing: all that efficiency will start to erode if you need to create a new copy of a given course for every language on your books.

That’s why Gomo allows you to pair the hands-off convenience of AI translation with multiple language layers. When you create a course, you’ll be presented with an option for multiple languages: a simple process in which you find the languages you need and add them to your course. For every language you select, you’ll gain a new language layer, allowing you to effortlessly populate each layer with AI-powered translations—all within a single course.

By combining AI translation with multiple language layers, you’ll preserve the speed and effectiveness at the heart of your translation tech, catering to a huge range of learners in the blink of an eye.

Why human translation (and XLIFF exports) still have a part to play in eLearning translation

Translation might be an impressive use case for ever-advancing AI capabilities, but (as with all things artificial intelligence) that doesn’t mean the human element isn’t still an essential piece of the translation puzzle.

While AI translation tools absolutely can do a fantastic job at producing sophisticated eLearning translations, there’s always a case to be made for consistent human oversight. And there are plenty of specific occasions when you might want a human translator to get involved in the process—or to take over entirely. For example:

• If you’re producing eLearning content that deals with complex, technical, or highly specialized topics, an appropriately specialized human translator might be a better choice for handling translations for that content.
• If you’re putting together courses that come with ethical or sensitive connotations, you might prefer to hand over any translations to a human who can treat those topics with the careful touch they need.
• If you’re happy to let AI take care of the brunt of your translation needs, you might still want a human eye on the end result for quality assurance purposes, either as a consistent hand on the wheel or in a spot-checking capacity. The latter doesn’t necessarily require a full XLIFF export—but Gomo’s cloud-based collaboration features make it simple for human translators to dip in and out of AI-produced content as often as you see fit.

In these scenarios, you’ll want to make sure that you still have the means to pass on your content to human translators or agencies. That way, you’ll supplement the convenience of automation with the extra layer of quality and oversight that a human touch brings to the table.

Fun fact: Gomo offers XLIFF exporting and content review capabilities alongside its AI translation feature

Gomo’s all about keeping things simple. That’s why we’re known for our pre-built visual themes and rapid delivery options. So, to ensure your learning content is never more than a few clicks away from a translation agency, Gomo makes it easy to export XLIFF translation files.

At the press of a button, use Gomo to create a zip file containing everything you’ll want your translators to translate—including little details like the text on buttons or in pop-up messages that learners might encounter.

Gomo will also ensure you can review your AI-translated content with collaboration features that make it simple to assess content and leave feedback where appropriate.

Keeping your data secure when using AI

Your eLearning content is a form of data—and, as with any form of information your organization holds, that data needs to be treated securely.

This is a subject that’s especially worth bearing in mind when it comes to AI. With the advent of generative AI tools that sometimes train themselves on data supplied by users, it’s never been more important to ensure that you know what’s happening with the information you’re sharing.

That’s why Gomo’s AI translation feature is powered by our customers’ own OpenAI accounts. Not only does this allow them access to some of the most powerful AI around, but—more importantly—it ensures that Gomo doesn’t send information anywhere except to your own OpenAI account.

A final word on AI translation

Human nuance will always have a part to play when it comes to translation. That’s why, in the land of Gomo, our XLIFF exporting and robust content review features aren’t going anywhere.

But when you need translations that combine the naturalistic style of human writing with the scaled-up efficiency that only automation can provide, there’s no question that AI translation technology is the perfect way to fill that gap—and its capabilities are only going to grow.