16 years after Google Translate, has artificial intelligence replaced translators?
While recent advances in artificial intelligence have worried writers and artists, the translation industry has been affected by this problem for much longer.
Artificial intelligence has been the star of tech news in recent months as they have made impressive progress. Advances that have also caused concern and anger, as they raise the issue of humans being replaced by robots and artificial intelligence. The translation industry has been struggling with these questions for several decades.
Technologies and the labor market are also developing
For a long time, machine translation software like Google Translate was derided for being so verbose and incapable of grasping context. This changed in the 2010s with the advent of neural networks, which are able to process entire sentences at once to create smoother and finer translations based on context.
The most popular neural machine translation software – because it’s free – is DeepL. announced himself “The best translator in the world” on its website DeepL is a multilingual dictionary created in 2017 by the German company Linguee, known for its website of the same name, which allows you to find the translation of a word or phrase by comparing texts in both languages. That’s why DeepL relies on this first site’s database to make more accurate translations.
This has partly changed the work of freelance translators, as clients sometimes approach them with automatically translated texts and offer them only proofreading and editing. A faster and cheaper solution for clients, but frustrating for translators because sometimes whole paragraphs have to be redone for less. Because translation software is not flawless, its ubiquity leaves translators with the bitter impression that their clients prefer quantity over quality.
Work around limitations
Having been interested in machine translation tools for a long time, I regularly try to push them to their limits with a common French expression: “I drank a cup”. Even a child knows that, depending on the context, this phrase can mean that I accidentally swallowed water while swimming. However, advanced machine translation software like DeepL can’t literally translate it for me, even if I add the context of swimming. It’s really an idiom that doesn’t necessarily have an equivalent in another language, so Linguee’s comparative text base doesn’t help it at all. When the swimming context is defined, it is available by clicking on it he drank (it) to obtain many alternative translations. Among them, suffocated (to drown). Therefore, human intervention is required to steer DeepL in the right direction. The site then prompts me to add it to my dictionary to drink translated by suffocate but not to capture the entire expression.
“The examples given so far seem harmless, but in critical contexts where the translation must be completely accurate, such as in diplomacy, the slightest mistake can have serious consequences. For literature and audiovisuals, people will always be more adept at conveying a certain style, rhythm or emotion. »
For some languages, the software also has workarounds with English, as it is the language with the most comprehensive database. This sometimes leads to strange translation errors, as noted by teacher-researcher Pascal Elbaz last May in “Will Neural Machine Translation Replace Humans?” explained at the conference. » : he cites a text in Chinese about a calligrapher who also likes to carve seals. Sealit is in english stamp. Sealcan always mean in English as well stamp. Therefore, the French translation describes a calligrapher who likes to carve seals.
The examples given so far seem harmless, but in critical contexts where the translation must be completely accurate, such as in diplomacy, the slightest mistake can have serious consequences. For literature and audiovisuals, people will always be more adept at conveying a certain style, rhythm or emotion. Therefore, human translators remain indispensable for many reasons.
On the plus side, artificial intelligence can be an ally for these translators. This is called CAT, computer-assisted translation. This software allows them to maintain stylistic and terminological consistency, for example by memorizing translations of specific phrases. While translators lament that prices are being driven down by AI, they cannot be replaced and may even become more efficient thanks to these technologies.
