Will ChatGPT Take Over Translation Services?


Unless you’ve been living under a rock over the last few months (or don’t spend every free moment on social media), you’ll have seen endless tweets and think-pieces about the wonders of ChatGPT. Since it was opened up to the public in November 2022, users have been shocked at the accuracy, and have been flocking to similar tools to test out their reliability and effectiveness — particularly in the digital marketing arena.

For us, using AI is nothing new — in fact, we’ve been using natural language processing and machine translation software in various forms for years. However, as AI writing tools and bots become more widespread, will we rely on them even more? Will they be able to produce content that appeals to users and overtake copywriters, translators and content marketers?

Read on to find out more…and forgive us for adding to the already crowded “ChatGPT” content space!

What is Natural Language Processing?

Natural language processing, or NLP, combines computer science, linguistics and artificial intelligence to teach computers to understand and process human language. The overall goal is a computer that is capable of not just “understanding” content but is also able to interpret the contextual nuances of the language used. 

NLP breaks down the text into bitesize chunks, and with the help of machine learning algorithms and statistical models, the computer can learn the patterns of the inputted language and create its own output based on this learning.

So that’s the (basic) how, but why?

Put simply, delegating typically monotonous or repetitive tasks to a natural language processor increases efficiency. Whether it’s a customer service chatbot on an eCommerce site, Siri or Google Maps giving you directions, or even a program that pulls out medical information from doctor’s notes, there are endless examples of how NLP can be useful.  

For digital marketers, NLP has many uses:

  • Sentiment analysis – helps brands understand how customers are responding to their products/brand messaging based on online reviews or social mentions.
  • SEO strategy – NLP can identify the keywords being used so brands can utilise them in their content strategy.
  • Audience analytics – NLP can access data from various online platforms so brands can reach their target audience and serve them with more relevant paid ads. 
  • Lead generation – a chatbot on a homepage can act as an auto-responder for basic lead qualifiers.
Lines and lines of multicoloured HTML code depicting chatgpt

Natural Language Processing and AI in Translation

If you have any involvement in translation or transcreation, you may already be aware of machine translation (MT) tools. They employ NLP to understand the inputted text in order to reproduce it in the new language. 

Google Translate is perhaps the most famous — enabling anyone to exhibit rudimentary language skills with a quick click of a button. Over the years, Google has expanded and refined its linguistic capabilities, with over 103 languages in the database as of January 2023. By employing neural machine translation, the engine attempts to translate sentences all in one go, rather than word by word, by taking context and grammatical clues into account. 

But, despite Google’s immense power, there are still examples of the Translate service not quite nailing the brief. For example, the Finnish idiom “menee yli hilseen” directly translates to “goes over the dandruff”, but colloquially means “too hard to understand”. So although it may be enough for simple, conversational translations, it won’t hit the mark for longer form content — especially if it’s to entice customers to make a purchase. In fact, it’s likely to alienate them more than using English! 

Google Translate image showing Finnish: Menee yli hilseen Actual meaning: Too hard to understand

Finnish: Menee yli hilseen

Actual meaning: Too hard to understand

Google translate showing Portuguese phrase "Dor de cotovelo" & English translation "Elbow pain"

Portuguese: Dor de cotovelo

Actual meaning: To be jealous

Google Translate image showing Hungarian: Kutyából nem lesz szalonna! Actual meaning: You won’t become something you’re not

Hungarian: Kutyából nem lesz szalonna!

Actual meaning: You won’t become something you’re not

Currently, any MT output needs a human to edit, tweak and reword to make sure it’s not only accurate but as engaging and contextually aware as possible. And this is really the crux of the conversation surrounding ChatGPT and other similar tools. Can these super-sonic AI chatbots really mimic the human element that makes marketing content effective and engaging?

Transcreation, Localization & AI

When it comes to translating marketing messaging effectively, we always advise our clients to consider transcreation services over translation. Transcreation takes the fundamental idea of the original copy but transforms it using cultural insights, local knowledge and a native understanding of the language, to ensure it appeals to the new target audience. 

But transcreation goes further than language. It can help shape entire marketing campaigns, alter promotions and even impact product launches. Having “feet on the ground” in the new target market can give brands valuable insight into the region and culture that no amount of online research can provide.

Do you see where we’re going with this?

ChatGPT, and all the other AI tools out there, rely solely on the information they’re given. Of course, there’s a wealth of information online about different cultures, religions and news stories that the tools can utilise. But is that enough to truly understand the world around us? Social media trends and influencers are constantly in flux, as is language itself. A brand targeting 20-somethings will need to mimic their phrases to attract customers, and it won’t be the same vocabulary needed for a brand targeting the over 60’s, that’s for sure! 

It’s this cultural and contextual nuance that can make or break a marketing campaign, even a brand. We’ve all seen what happens to culturally, or socially, insensitive campaigns — they’re instantly panned across Twitter causing the brand to scramble wildly to create an alternative ad or knock up a quick apology. 

There are many examples of ChatGPT tripping up when faced with basic logic problems, so being able to delve into complex cultural or religious details will be far beyond it’s reach for a long, long time. 

This isn’t to say we’re not fans! There are huge benefits to using such a tool, and we’re keen to see how it evolves in the future. But for now, we’ll stick to our talented bunch of linguists to get the very best work. 

We haven’t covered the SEO implications of using these tools, which is a whole other kettle of fish. Come back for Part 2!

Get in touch to find out how we can take your brand global!