Translation & Localisation

Machine translation has improved enormously, and for many common language pairs and straightforward content, AI translation is now genuinely useful. Tools like Google Translate and DeepL produce translations that are often good enough for everyday comprehension and basic business communication. For more demanding use cases - legal documents, marketing copy, literary texts, or content requiring cultural adaptation - AI serves as a powerful first draft that human translators refine, significantly speeding up the process. The biggest gains have been in language pairs with abundant training data, particularly between major European languages and English. Translation for lower-resource languages - many African, Asian, and Indigenous languages - remains significantly weaker, reflecting the training data these models were built on. Localisation goes beyond translation: adapting content for cultural context, local conventions, humour, and tone. This remains an area where human judgement is essential. AI can handle the mechanical aspects of translation well, but understanding that a joke does not land in another culture, or that a colour carries different connotations, requires cultural fluency that current AI systems simply do not possess.