What Content Marketers Can Learn From the Translation Industry About Using AI

Thirty years ago, localization expert Jean-Luc Saillard introduced computer-assisted translation (CAT) tools to the translation agency he was working with at the time. CAT was a pre-AI technology that helped automate the translation process.

It didn’t go well.

“Translators worried that relying on translation memory would make their work more repetitive, and decrease the pay and amount of work available. Twenty-five years later, the same arguments are being made about machine translation,” he writes.

For over thirty years, there’s been a strong contingent in the language service provider (LSP) community who believe that relying on technology will harm the quality of translations. 

They’re not wrong. While AI-enabled translation has come a long way, no translation tool perfectly understands the nuance of context. And when unchecked, machine translations can lead to sometimes humorous — sometimes disastrous — misunderstandings.

Last year, a news site warned Germans against buying hamsters due to the ongoing food shortage. While in a 2020 machine translation mishap, Facebook translated Chinese President Xi Jinping’s name to “President S***hole.”

So yes, in 2023, the translation industry still requires a human touch: translators and editors capable of reading context and identifying when something seems off, not to mention the world of translation that is in itself an art form. Would you want to read poetry translated by a computer? (Don’t answer that.)

But let’s talk about sustainability. According to Jean-Luc, translation businesses that are not utilizing machine translation in one form or another are not able to compete with those that have. “The reality is that machine translation is cheaper, faster, more secure, and increasingly better quality,” Jean-Luc explains. “LSPs that do not adopt this quickly dominating technology will not be able to compete in this new market.”

There are cost and time considerations. Companies that utilize machine translation are cheaper with faster turnarounds  — even when you factor in human post-editing. And especially for larger projects, faster and cheaper is usually going to win out.

Machine translation can actually increase quality, as well. Utilizing translation memory — a database of text and terminology and preferred translations — in coordination with machine translation can increase the consistency of a project or client-specific lexicon. Similar to tools like spellcheck and Grammarly, machine translation tools can often catch more errors than a human can. 

All in all, the localization industry is stronger because of AI. And the same is going to be true about marketing ... eventually.

Buzzfeed just announced it would begin using OpenAI, the creator of ChatGPT, to create personality quizzes and assist staff in personalizing content. Not surprisingly, Buzzfeed’s shares jumped more than 150% after this announcement.

Marketing and media organizations are hungry to jump on the #ChatGPT bandwagon. But my fear is that we are going to misuse these tools in order to simply crank out more low-quality content. And it’ll be easy to do. A number of AI-enabled tools are cropping up that promise high-quality computer-generated content on supposedly any topic.

This, of course, will only add to the content pollution I wrote about last week. Even if artificial intelligence is capable of writing content that’s indistinguishable from that written by a human, is it doing so by simply mimicking much of low-quality content that’s already out there?

While this democratizes content creation by making it easier for any business to produce content at lower price points, it also will make it significantly harder to stand out with unique ideas or find resources from verifiable experts.

Meanwhile, there are better uses of the technology that can enhance the performance of media and marketing professionals while making our jobs easier.

Many of the same tools that write entire posts for you come with other features that can be useful even for AI critics like myself:

1. Generate ideas for blog posts, keywords to target, ad copy.

2. Write first drafts of press releases, blog outlines, etc.

3. Generate metadata.

4. Recommend clips from longer videos or quotes from long-form posts.

5. Make transcripts more useful with automated outlines and topic clusters.

In order to stay relevant, every multimedia creator, content strategist, and content manager needs to begin forming a strategy for how to incorporate AI technology into their workflows.

And we certainly won’t be the first industry to have done this work.

Over the years, the localization industry has learned smart ways to integrate machine translation into the translation process. For instance, in his post, Jean-Luc identifies situations that call for different levels of post-editing. Little or no human post-editing may be needed for translating support forums or internal communication in a company. He also believes there are situations in which machine translation will not work, such as public-facing materials that require enhanced cultural sensitivity, such as marketing content.

Similar to what I said last week, as we explore the uses of new technology, I hope we keep in mind that whatever we produce is how the public will shape its opinion of our brand. Do we want that opinion to be based on content created by a computer or a human? 

Featured image Image by Mohamed Hassan.