Professor Mahmoud Kamel
Artificial intelligence is bringing major changes to the localisation industry. This is not simply another technological development or a new tool added to existing systems. AI is changing the way content is prepared, translated and adapted for international markets. It is also changing the work of localisation professionals, the way companies organise their services and the skills needed in the profession.
The main question today is no longer whether AI will affect localisation. It is already doing so. The more important question is how far this change will go and how localisation professionals can adapt to it. This is the main question discussed in the third chapter of Translation Studies in the Age of Artificial Intelligence, edited by Sanjun Sun, Kanglong Liu and Riccardo Moratto. The chapter, written by Nick Lambson and Lintao Han, is entitled “Localisation in the AI Era”.
The changing face of localisation
Localisation has always involved more than translation. It means adapting products, services and content to meet the linguistic, cultural and technical needs of different markets. Traditionally, localisation professionals have tried to achieve a balance between three main factors: speed, cost and quality.
Today, AI is changing this balance.
The amount of content that needs to be localised has increased greatly. Digital platforms, international trade and the growing demand for information in people’s own languages have all contributed to this increase. Traditional localisation methods, which depend mainly on human translators and established workflows, are finding it difficult to deal with this growing volume.
Neural machine translation has become more advanced and, in some cases, can produce results close to human translation. Other technologies, such as quality estimation and automatic post-editing, are also helping to improve speed and efficiency.
AI is also making localisation more multimodal. Localisation professionals are no longer dealing only with written texts. They may also work with images, audio, video and interactive content. As a result, the boundaries between translation, content creation and multimedia production are becoming less clear.
Some specialists describe this development as a move towards a “post-localisation” era. New expressions, such as “Language Data for AI” and “Language Operations”, have also appeared. These terms reflect the growing role of language throughout the whole process of creating and distributing content.
In the past, content was often produced in one language and then translated into others. Today, multilingual content may be created from the beginning, sometimes with the help of AI.
How AI is changing localisation workflows
AI is affecting almost every stage of the localisation process.
One important area is source text analysis. Large language models can now analyse a text and produce reports about its style, tone and discourse features. This information can then be included in translation instructions or prompts to help produce better translations.
This can help overcome one of the traditional weaknesses of machine translation: the tendency to produce correct but flat and general language that does not always reflect the style and voice of the original text.
Content classification is also becoming easier. AI systems can analyse a text and decide which translation method may be most suitable. For example, one translation engine may perform well with product descriptions but may not be suitable for medical or technical content.
AI can, therefore, help organisations decide whether a text should be translated automatically, post-edited by a human translator or translated completely by a professional translator. This allows human translators to focus more on complex and important texts where their expertise is most needed.
Machine translation itself is also changing. Neural machine translation remains widely used in professional localisation, but large language models are becoming attractive because of their lower cost in some situations.
However, lower cost does not always mean better performance. Organisations still need to consider speed, consistency, quality and the ability of AI systems to work with existing localisation platforms.
Another important issue is language bias.
Many large language models have been trained on internet data, where English represents a very large part of the available content. As a result, their performance may be stronger in English and other widely represented languages than in languages with a smaller digital presence.
This remains an important challenge for companies working in multilingual environments.
New roles and new tools
The spread of AI is also changing the work of localisation professionals.
Computer-assisted translation tools are becoming more intelligent. They are combining translation memories, terminology databases, machine translation and AI models in one working environment.
This allows translators to receive suggestions based not only on the current sentence but also on previous translations, terminology and stylistic preferences.
The role of the project manager is also changing.
AI can help project managers plan projects, classify content and make decisions about the most suitable translation process. Administrative tasks can also be automated. This gives project managers more time to focus on communication with clients and translators and on the general strategy of localisation projects.
Localisation engineers are also playing an increasingly important role.
They are using AI to deal with technical challenges and large volumes of content. AI can help with tasks such as content segmentation, quality assurance and workflow automation.
The growing use of large language models also means that localisation engineers need new skills. In addition to programming and systems knowledge, they may need to understand AI systems, prompt engineering, natural language processing and workflow design.
Language data for AI and language operations
One of the most important developments is the growing demand for language data.
Language service providers are increasingly receiving requests for data collection, annotation and validation. AI systems need large amounts of high-quality language data, and language professionals can play an important role in preparing and checking this data.
This means that language companies are no longer dealing only with translation. They are also becoming involved in providing the linguistic resources needed to train and improve AI systems.
Another new concept is “Language Operations”.
The idea is to treat language services as an important part of an organisation’s general operations rather than as a separate activity that takes place after a product or service has been created.
Supporters of this approach believe that language should be considered from the beginning of product development, marketing and customer service.
However, not everyone is convinced.
Some specialists believe that Language Operations is simply another fashionable term. They point out that organisations have been trying to centralise and coordinate language services for many years, often facing problems caused by complex structures and different types of content.
Nevertheless, the idea has helped draw attention to the strategic importance of language in an AI-driven world.
AI in practice
The use of AI can already be seen in major international companies.
Duolingo, for example, uses AI in different parts of its language-learning system. AI can help create learning content, analyse learners’ progress and suggest exercises suitable for different levels.
The company also uses a human-in-the-loop approach, where human learning designers continue to play an important role in checking and improving the content.
Baidu has also developed systems that combine AI with human expertise. Initial translations can be produced by machine translation or large language models and then sent to translators who are considered suitable for the task.
The system can learn from human corrections and remember users’ preferred terminology and style. AI can also explain translation suggestions and offer recommendations for proofreading and improvement.
These examples show that the most useful applications of AI are often those that combine technology with human knowledge.
Human expertise still matters
Despite the rapid development of AI, the future of localisation is unlikely to be based on replacing human professionals completely.
AI can perform repetitive tasks quickly and can process large amounts of information. It can also provide suggestions and identify patterns that may be difficult for humans to detect.
However, localisation still requires creativity, cultural understanding and careful judgment.
A translation may be linguistically correct but culturally unsuitable. An AI system may also fail to understand irony, humour, political sensitivity or the expectations of a particular audience.
For this reason, the relationship between humans and AI should be seen as co-operation rather than competition.
Localisation professionals are increasingly becoming managers of AI-human workflows. They need to know how to use AI tools effectively, how to write useful prompts and, more importantly, how to evaluate AI-generated results.
Critical thinking will, therefore, become even more important.
The future professional will not simply accept what AI produces. He or she will need to decide when AI can be trusted and when human intervention is necessary.
The idea is gradually moving from “human in the loop” to “human at the core”. Human professionals should not only correct AI mistakes after they occur. They should also guide the use of AI, set standards and make the final decisions about quality and cultural appropriateness.
Looking ahead
AI is clearly opening a new chapter in the history of localisation.
It offers important opportunities for faster work, lower costs and the ability to deal with much larger amounts of multilingual content. At the same time, it raises serious questions about data privacy, bias, quality and the ethical use of AI-generated content.
The localisation industry will therefore need to develop new skills and new working methods.
Future localisation professionals will need strong language skills, but language ability alone may no longer be enough. They will also need technological knowledge, adaptability and the ability to work effectively with AI systems.
The profession is becoming more interdisciplinary, combining language, technology, data and human judgment.
Yet the basic purpose of localisation remains the same: helping people in different parts of the world access products, information and services in a language and form that meet their needs.
AI can become a powerful partner in achieving this goal. But technology should support human expertise, not remove the need for it.
The future of localisation will depend on finding the right balance between the speed and efficiency of AI and the cultural understanding, creativity and judgment of human professionals.
Those who can successfully combine these two sides will be in the best position to shape the future of localisation in the age of artificial intelligence.
By Dr Mahmoud Kamel
Professor at the Academy of Arts
