AI series, part 3/3: How to implement and make use of artificial intelligence in your translation and localization projects

AI series, part 3/3: How to implement and make use of artificial intelligence in your translation and localization projects

AI series, part 3/3: How to implement and make use of artificial intelligence in your translation and localization projects 1920 1465 Bartosz Budzynski

In this three-part series, we discuss the use of artificial intelligence in translation and localization. In our first post, we looked at how AI is altering the face of the industry. In our second post, we examined ways to boost your translation projects by harnessing AI. In this third and final post, we provide practical advice and some top tips on how to integrate AI into your translation and localization workflows.

1. Harness augmented and adaptive MT translation – where appropriate

In localization and translation, the use of artificial intelligence is most visible in machine translation. And this technology is changing and improving all the time. Augmented and adaptive machine translation are becoming increasingly widespread. They offer huge advantages – most notably, allowing translators and localizers to work more efficiently. So it makes sense to make the most of what AI can offer here. What’s more, there are now more sophisticated evaluation mechanisms that can help you better judge MT quality. Edit distance evaluation, for instance, allows you to assess the quality of MT output by looking at the number of changes a reviewer makes.

But remember that MT is not the be-all and end-all. The nature of MT means it works wonderfully well for certain text types, in certain scenarios. But not others. Think of marketing texts or highly creative texts that require a transcreative approach. AI doesn’t fit the bill for this type of work. It can’t create a text that draws on in-depth and up-to-date cultural references and knowledge of the target market, or that makes use of puns or rhetorical devices such as alliteration or parallelism for added impact. At least, not yet.

So, yes, use AI in translation. But where appropriate. Take an analytical approach rather than applying it across the board.

2. Automate your workflows – or parts of them

As we discussed in our second post, increasing numbers of translation providers use workflow automation to save them time and money. But here too, we recommend thinking carefully about what will work best for each client and project type. Full automation won’t be the way to go for many. But you’ll probably want some tasks to always be automated, such as file transfer, PO or invoice generation.

Similarly, consider whether you want to automate vendor selection. What if you had more options here, with the ability to get more granular in terms of specializations and text types, allowing you to make better matches? This is another area we predict will grow in the months to come. So keep your eyes peeled for changes, and think about how you can make the most of them.

3. Make your vendors’ lives easier

Use AI to help your vendors out, extracting key information from client documents and data, and providing your vendors with more context and terminology.
You can also make use of AI’s data processing abilities to spot issues with texts, such as inconsistencies and typos, and highlight these in the CAT tool in advance, saving your vendors time.

This will empower your vendors to create better texts – it’s a win-win.

4. Use AI to stand out from the crowd

With artificial intelligence developing constantly and being applied in new ways all the time, the possibilities are almost endless. AI can already help you create more accessible texts, for instance through automatic alt image text generation. It can also help you adapt a text to a particular style and tone of voice and advise you on how to optimize content for SEO purposes. It seems logical that soon it will be able to help you write more inclusive texts. In fact, Grammarly Premium already alerts users about the non-inclusive language they’ve used.

The next step is to take these tools and tailor them to the translation and localization workflow and requirements, perhaps embedding some into your CAT tools.

Our advice? Keep up-to-date with these developments and capitalize on any opportunities that can help you create better texts.

Three top tips for implementing AI into your translation and localization projects


1. Prioritize

Don’t try to implement AI everywhere all at once. Focus on your company’s goals and pain points and think about how AI might be able to help in those areas.

2. Communicate

Understandably, some translators – and internal staff such as project managers, for that matter – can get a little jumpy at the mention of AI or MT. It’s natural. No one likes to feel their livelihood is under threat. So it’s important you explain how you’re implementing AI and why. Explain it’s not to replace them, but rather to allow them to work faster, and focus on more fulfilling work.

Encourage your internal and external team to be a part of the process too. Because your vendors and PM team are perfectly placed to tell you which aspects of their workflows are menial or repetitive and could be automated, and which need the human touch.

3. Adapt

AI is changing all the time. For one, natural language processing is making great strides in enabling computers to understand ‘natural’ human language. As it continues to evolve, keep up to date with the changes and new capabilities. This way you’ll be able to spot opportunities and new ways AI can help your business, and adapt accordingly.

Bartosz Budzynski

Bartosz Budzynski

Responsible for Professional Services at XTRF. A strong supporter of the open-source movement, sharing economy enthusiast, and a passionate developer. His greatest superpower is transforming business needs into technical requirements.

All stories by : Bartosz Budzynski