How can we strike the balance between human and machine translation?
Where is that sweet spot where technology assists us without interfering with the review process?
How can we use technology alongside linguists without it overshadowing the human touch that’s so necessary in many translations?
How can we streamline the process with AI (Artificial Intelligence) without sacrificing quality or natural expression?
The list of questions could go on. The challenge lies in finding the way to harness the synergy between technology (AI) and linguist (post-editing) to achieve the most optimal result.
At LocalizationLab, this is an issue are at the forefront of our discussions for two main reasons:
Firstly, we aim to work with a team of professional linguists and create an environment where they thrive, ensuring they don’t have to abandon their careers due to technological advancements. And secondly, our goal is to integrate AI in a way that enhances our work, to guarantee that our translations are always excellent quality.
Technology allows us to speed up translations, which is sometimes crucial for projects that need to launch simultaneously worldwide, or in emergency situations that require immediate communication, for example. It helps us accelerate processes and saves time. However, relying solely on AI-generated results can lead to errors, ranging from minor nuances to serious issues that alter the original meaning.
Technology can help us to detect unnecessary changes in terminology (inconsistencies) by using well-defined glossaries. But when a term has multiple meanings, it struggles to determine the correct context and appropriate usage each time.
© HP Inc.
Example of a term with various meanings: In English, both images above can be referred to as “folders”. In Catalan, the words are completely different: “plegadora” (for paper), versus “carpeta” (for saving electronic documents). The work of a linguist is required to check the correct use of language in context.
Linguists, if necessary, can revisit the original content multiple times to determine the right translation. Depending on multiple factors, they may decide to use a more empathetic or humorous tone, if appropriate. Empathy, humour, cultural adaptation (localization), the correct tone for each brand, and creativity are all areas where AI in translation still falls short.
AI is evolving rapidly, but still struggles with certain language pairings. And it’s not terribly good at highly specialised texts either.
AI generated image by LocalizationLab via Canva
Wikipedia defines Yin and Yang as “a cosmological principle that explains the duality of everything that exists in the universe: The One or Tao (or Dao) divides into two fundamental forces, opposing but complementary, which are at the origin of all things.”
For us, AI and the work of our linguists are akin to Yin and Yang – two complementary forces that must learn to coexist and respect each other. We continually strive to find the balance to make processes efficient and cost-effective, while ensuring the results are of high quality.
Before starting a translation project, we recommend seeking the advice of specialists to determine which tools to use in each case (TAO, LLMs, Translation Memories, glossaries, etc.). It is also important to check the level of security and privacy protection of the tools, depending on the information you want to translate.
At LocalizationLab we always aim to be at the cutting-edge of the technologies and processes that help us achieve excellent quality translations using all the resources available.
(*) Remember, oxymorons are a rhetorical figure that unites two concepts with opposite meanings. Some examples: holy war, clear night, organised chaos, awfully good.