Google's translation service handles one trillion words per month. The French say, "Il pleut des cordes," which translates directly to, "It's raining ropes." This idiom can confuse AI translators. François Yvon, a research director at the French National Centre for Scientific Research (CNRS), explains, "You can imagine the dilemma this throws AI into." A well-trained system recognizes that ropes are not literally falling from the sky and would use the English phrase, "it's raining cats and dogs." However, Yvon questions whether it can find an equivalent when translating into Indonesian.
The Rise of Machine Translation
As translation increasingly shifts to machines, the stakes are rising. In April, Google reported that users are translating an external one trillion words monthly with Google Translate. This figure excludes other platforms like DeepL, Microsoft, Claude, ChatGPT, and Gemini, as well as custom corporate translation systems. The direct sales of machine translation outside general AI subscriptions were valued at an estimated $1.4 billion (£1 billion) last year.
John Zimmerman, founder of the Open Knowledge Association (OKA), states, "For plenty of use cases, raw AI translation is now the right call." He argues that multiplying costs to catch the last few percent of errors isn't rational for internal documents or product listings. However, he notes that for an encyclopedia like Wikipedia, accuracy is paramount. "A wrong fact can sit there for years and get cited by journalists and AI systems," he adds.
With over 10,000 published pieces translated primarily into Spanish, Portuguese, and English, OKA is the largest contributor of translated articles to Wikipedia. Zimmerman explains that AI plays a crucial role in this process: "Translating 10,000 articles the traditional way would cost millions. In our process, the AI produces the draft, then a human translator checks every sentence."
