How AI is Democratising Translation Services

On November 30th, 2022, the world changed forever. On that day, OpenAI launched its text-generating AI chatbot, ChatGPT, and the public’s understanding of artificial intelligence (AI) was fundamentally transformed. Although this new, exciting technology seemed like an overnight invention, many industries had actually been using AI and machine learning for decades. 

In fact, the journey of machine translation (MT) has deep roots in the translation and localisation industry, beginning as early as the 1950s. One of the first major milestones was the Georgetown-IBM experiment in 1954, where a computer was trained to translate Russian sentences into English during the Cold War. 

Then in 2016, Google revolutionised MT with the introduction of Neural Machine Translation (NMT), a new technology that led to wider adoption and improved translation quality across various MT tools. 

With the inclusion of translation apps on mobile phones, access to automatic translations has been available with the press of a button for several years. Combine this with the growing number of GenAI tools (ChatGPT, Gemini, Llama, etc.), and the democratisation of translation services is truly reaching its golden age.  But what has driven the rapid rise of this technology? What opportunities does it present for businesses today, and what could the future of AI translation hold? Let’s explore those answers and more.

The Driving Forces Behind AI in Translation

Why have tech giants and industries invested so heavily in developing MT, NMT and large language models (LLMs)?

The answer lies in the explosion of content. In today’s world, content is king, and it’s being generated everywhere, by everyone. From marketing materials to product documentation or social media posts, this content demands rapid, accurate translation across numerous languages to serve global audiences. The increasing demand for multilingual content was clearly illustrated by TAUS’s 2016 Evolution of the Translation Technology Landscape. At the end of the 20th century, most content was localised into two or three languages. Fast forward just over 20 years and that number has dramatically increased. Only a small fraction of this vast amount of content can be handled by human translators alone. Automated translation is therefore essential to make this growing volume of information accessible on a global scale.

Image source: TAUS 2.0 and the Game Changers in Localization, by Jaap van der Meer

How AI Has Made Translation Democratic

For decades, professional translators have used Computer-Assisted Translation (CAT) tools to help with their work. Machine Translation (MT) has been integrated into these tools for some time, allowing professionals to use these technologies. Meanwhile, automated translation services and apps have become accessible to the general public.

Tools like Google Translate and DeepL have made it easier for individuals and companies alike to translate content quickly and at little or no cost – although it is important to note that these tools come with their own serious quality and security risks. This democratisation of translation services has unlocked new possibilities for individuals to consume global content in their own language and for businesses to reach international audiences.

For decades, professional translators have used Computer-Assisted Translation (CAT) tools to help with their work. Machine Translation (MT) has been integrated into these tools for some time, allowing professionals to use these technologies. Meanwhile, automated translation services and apps have become accessible to the general public.

Tools like Google Translate and DeepL have made it easier for individuals and companies alike to translate content quickly and at little or no cost – although it is important to note that these tools come with their own serious quality and security risks. This democratisation of translation services has unlocked new possibilities for individuals to consume global content in their own language and for businesses to reach international audiences.

Case Study: eBay’s Use of AI for Cross-border Trade

A great example of AI democratising translation is eBay’s use of AI-powered translation to enhance cross-border sales. In 2014, eBay launched its eBay Machine Translation (eMT) system to break down language barriers for buyers and sellers around the world.

eBay didn’t translate the content itself but provided a tool that allowed buyers to read product listings in their native language. This shows how important it is for buyers to have access to information in their language to make informed purchasing decisions.

According to a study by MIT and Washington University, the adoption of eMT led to an 11% increase in sales from the US to Spanish-speaking countries. The tool improved the accuracy of product titles and search queries, helping buyers understand listings more clearly, which in turn drove sales.

Opportunities for Businesses

While businesses like eBay use AI for cross-border trade, this technology has also given companies more options for how they use translation tools:

● Understanding the gist

For everyday communication or simple tasks, where precision isn’t essential, machine translation tools can provide a quick way to understand the general meaning of a text. This is often sufficient for understanding internal emails, social media posts, or other informal exchanges.

● Internal post-editing

For businesses with in-country resources or employees fluent in multiple languages, machine-translated content can be post-edited internally. This allows the company to ensure that brand terminology, tone of voice and consistency are maintained while speeding up the process. When the right internal capacity is available, this approach helps companies fine-tune MT output to align with their brand and industry-specific needs.

● Professional post-editing

However, for companies lacking internal linguistic expertise or resources willing to take on extra linguistic tasks, professional post-editing becomes crucial. In this case, language service providers can step in to guarantee that the machine-translated content is refined to meet quality standards, ensuring that it aligns with the company’s terminology and overall brand message.

● Professional translation services

For high-value content, such as legal documents, marketing campaigns, or technical manuals, relying on human experts remains essential. Human translators, supported by technology, ensure that translations are accurate, culturally relevant, and reflect the client’s voice and objectives.

With the rise of AI, especially NMT, companies no longer need to choose between translating everything or cutting costs. By balancing automation and human expertise, businesses can manage the challenges of global communication while staying within their localisation budgets. Language service providers have the expertise to help clients decide the best scenario and tools for their specific needs, ensuring they avoid mistakes and maximise their results.

Preserving Language Diversity

Another critical impact of AI in translation is its ability to preserve and expand access to lesser-known or low-resource languages. Machine translation tools are continually adding new languages, providing millions of people with access to digital content in their native tongues​​. This represents a massive leap forward for language diversity, which might otherwise be left behind in an increasingly globalised world.

However, the quality of MT varies significantly depending on the language pair. For widely spoken languages, the results can be quite accurate. Yet, for low-resource languages, human translation may still be necessary to ensure quality and work out to be both cheaper and faster.

Generative AI vs. NMT: Complementary Tools, Different Strengths

While both NMT and Generative AI are AI technologies, they serve different purposes. Generative AI is also transforming how linguistic tasks are approached. Possible GenAI use cases include term extraction, term injection, term base and translation memory TM clean-up, as well as automated evaluation of raw MT output. Automated quality checks, such as error spotting or consistency validation, are becoming more efficient thanks to GenAI. LLMS can also be used as the first step in post-editing, streamlining the process by identifying areas for human refinement.

Looking Ahead: The Future of AI and Translation

AI is poised to continue driving innovation in the translation industry. As NMT tools improve and generative AI becomes more integrated into translation workflows, businesses will enjoy even greater accessibility to global markets. While NMT remains the go-to technology for accurate, high-volume translation tasks, Generative AI could support more creative and flexible language use cases. For instance, in customer support scenarios, Generative AI could handle conversations in multiple languages, with NMT ensuring precise translation of technical queries.

The key to successful global communication lies in striking the right balance between leveraging AI for speed and affordability and relying on human expertise for precision and context. While AI tools make it easier to process large amounts of information and facilitate initial communication, human experts are still essential for quality, accuracy and cultural relevance. Machines help speed up the work but the process is far from fully automated; human input remains critical, especially for high-value content such as legal documents or contracts. Businesses must be informed and thoughtful in choosing the right combination of technology and expert oversight.

In our next article, we’ll take a closer look at the potential risks associated with machine translation and why human translators will continue to have a critical role in the industry.