How long will humans be needed for translation?

How long will humans still be needed for translation?

Google's translation service handles one trillion words per month

“Il pleut des cordes,” the French say when rain is falling heavily.

The literal English version is “It’s raining ropes,” which is exactly the sort of idiom that can confuse an AI translator.

“You can imagine the dilemma this throws AI into,” says François Yvon, research director at the French National Centre for Scientific Research CNRS, where he specialises in automatic translation.

A well-trained system, he says, will understand that ropes are not actually dropping from the sky and will choose the English equivalent, “it’s raining cats and dogs”.

“But is it going to find the equivalent when it’s translating into Indonesian?” asks Yvon.

Weather talk is relatively low-stakes, but machine translation is increasingly being used for subjects where the consequences are far more serious.

In April, Google said people are now translating one trillion words a month with Google Translate.

That figure does not include other services such as DeepL, Microsoft, Claude, ChatCGT and Gemini, nor custom corporate translation systems.

Direct sales of machine translation — meaning outside general AI subscriptions — were estimated to be worth $1.4bn (£1bn) last year.

AI translation is the “right call” in many cases, says John Zimmerman

“For plenty of use cases, raw AI translation is now the right call: multiplying your costs by 100 to catch the last few percent of errors isn’t rational if you’re translating internal documents or product listings,” says John Zimmerman, founder of OKA (Open Knowledge Association), a Swiss non-profit that translates Wikipedia articles into other languages.

“For an encyclopedia it’s different. Wikipedia runs on trust. A wrong fact can sit there for years and get cited by journalists and AI systems,” he says.

With more than 10,000 published pieces, mostly in Spanish, Portuguese and English, OKA is, Zimmerman says, the largest contributor of translated articles to the collaborative online encyclopedia.

He says that would have been impossible without AI.

“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,” he explains.

But that human role is shrinking.

“The machine output is now good enough that skipping the human entirely has become a serious option, and a real temptation across the industry. Two or three years ago, with tools like DeepL, our translators still rewrote a good chunk of every draft.

“Today the prose often doesn’t need fixing at all. So, the question is no longer whether AI translation is good, it clearly is, but whether it’s good enough to remove the human,” says Zimmerman.

Translators say not yet.

Ariella Baltié is a translator and legal specialist who works on court documents from English into French.

She was recently approached by someone who had already used AI to translate his documents and wanted her to review them.

“There were two major errors right at the beginning,” she said.

The AI translated “Le demandeur” (the plaintiff) as “the requester”.

“Totally awkward for legal papers,” says Baltié.

When the AI encountered “Madame X, épouse Y” (Ms X, wife of Y), it produced “Ms X marries Y”.

In other words, it treated a noun as a verb and completely lost the meaning.

The consequences of mistranslation in law can be severe. In the US, for instance, mistakes in machine-translated asylum applications have caused delays in detainees being released.

Ariella Baltié specialises in legal translation work

Accuracy can be a matter of life and death in medicine.

“The units for dosages and blood tests often include Greek letters,” says Carol Bereuter, a specialist in medical translations.

She encountered one automatically translated text that rendered the Greek letter μ as milli (one thousandth) instead of micro (one millionth).

“That’s a serious error. If you give someone a milligram when it’s meant to be a microgram, you are giving them a thousand times the dose,” Bereuter says.

She also notes that as AI translators improve, mistakes become harder to detect.

“Whereas in the past, with things like Google Translate, the sentence would jump out at you because there’d be a grammatical error or it would sound clunky, today it sounds perfectly legitimate. It sounds fluid. It sounds natural. So the errors are a lot harder to spot.”

Carol Bereuter is a specialist in medical translations

“If it’s something that will bear your signature, there’s a risk for your reputation. And if decisions are going to be taken on the basis of a translation concerning a medical procedure, a contract you might sign or a major purchase you might make, the risk is even greater,” warns François Yvon.

He says even sentences that seem straightforward to a person can confuse an AI service.

“When you have words with different meanings, it’s usually very clear for a human which one applies because they understand the context, but when a machine translator comes across the sentence ‘translate into Polish’, for example, it is quite capable of deciding it must be all about shoe cream,” says Yvon.

Context is often essential. If an AI translator has not understood that the subject is a football match, it may mistranslate “taking a corner”.

Last year OKA carried out a study funded by Wikimedia. It examined 119 articles, 10 language pairs and more than 1,000 hours of translator work.

“27% of the AI-generated text still got corrected before publication. Three quarters of those corrections were errors the AI introduced, and 5 to 6% of errors changed the meaning of the original text,” says Zimmerman.

Eliminating those mistakes is likely to be difficult.

“As things stand, I think the fundamental way these systems function makes it quite hard to imagine how they could become 100% reliable,” says François Yvon.

That is because probability is central to how AI chooses one word over another, and what is probably correct is not the same as what is correct, especially in something as subtle and unpredictable as language.

He believes a new generation of technology will improve matters.

“We will see translation systems that are increasingly capable of evaluating what they do and correcting themselves using different AI tools simultaneously: one to do the translation, a second to criticize it and a third to produce a new version, for example.”

But the major breakthrough may come when AI can express uncertainty.

“The real breakthrough won’t come until these systems are not merely capable of translating but also of telling us when they’re not sure,” Yvon says.