After completing his doctorate at the University of Tokyo, Tatsuya Amano shares a story about attending his first international scientific conference. Even after carrying textbooks around campus for years and studying English since seventh grade, he still found himself standing at a coffee break unable to follow half of the conversations going on around him. 2006 was that year. A researcher in his position might simply open an app almost twenty years later.
AI systems and human translators have been getting closer for years, but something changed sometime in 2024 or 2025. The deep-learning architecture of neural machine translation, which powers programs from Google, Meta, and OpenAI, started generating results that were difficult for experts to discern from those produced by qualified human linguists. AI translations were rated similarly to, and sometimes higher than, translations made by individuals with advanced degrees in linguistics in controlled tests involving widely spoken language pairs like English-German or English-Chinese. That’s not press release marketing language. Many people find it unsettling when it appears in industry benchmarks and peer-reviewed evaluations.
In short, machine translation is the major AI success story, according to Philipp Koehn, a professor of computer science at Johns Hopkins who has observed the field’s development for more than 20 years. It is practical and effective. From someone who remembers the cumbersome rule-based systems of the early 2000s, when linguists spent years manually encoding grammar and the results still read like furniture assembly instructions, there is a directness in that statement that feels earned.
However, there is a big difference between replacing bilingual people in the real world and outperforming them on a benchmark. The limitations come quickly, as anyone who has attempted to use AI to translate a legal contract, a poem, or even the slang-heavy dialogue of a racing video game will attest. Even outsourced human translators struggle without deep contextual knowledge, according to a Reddit user who translates technical manuals between structurally disparate languages. AI can’t even call an engineer to find out the meaning of a cryptic abbreviation. For any system that has been trained primarily on probability, the texture of language—its idioms, irony, and silences—remains stubbornly challenging.
The discussion seems to have split into two groups that aren’t really communicating with one another. For 90% of use cases, such as restaurant menus, support tickets, social media posts, and browser pop-ups that offer to convert a webpage, engineers and efficiency-minded executives view translation as a solved problem. However, multilingualism scholars, diplomats, and literary translators emphasize that comprehension and fluency are not the same thing. According to a survey of over 900 environmental scientists, non-native English speakers spent up to 91% more time reading papers and 51% more time writing them, and they were more than twice as likely to have their papers rejected for poor writing. AI could greatly reduce that burden. It’s a different and more difficult question as to whether it should take the place of the mental challenge of actually learning the language.
This is known as “desirable difficulties” in psychology. Cognitive resilience is strengthened by the challenges of finding the right word, struggling with unfamiliar grammar, and juggling two linguistic systems at once. Outsourcing all of that to a machine might save time while subtly undermining something we still don’t fully comprehend. The tension was succinctly summed up by Andy Benzo, president-elect of the American Translators Association: we are in a transition, but it won’t replace the translator.
It’s difficult to ignore the irony. The more proficient AI becomes at imitating human language, the more evident it becomes that language serves humans as a means of thought as well as a means of information exchange. The accuracy gap may have been closed by machines. However, we still own the meaning gap. For the time being.
London Bilingualism's content on health, medicine, and weight loss is solely meant for general educational and informational purposes. This website does not offer any diagnosis, treatment recommendations, or medical advice.
We consistently compile and disseminate the most recent information, findings, and advancements from the medical, health, and weight loss sectors. When content contains opinions, commentary, or viewpoints from professionals, industry leaders, or other people, it is published exactly as it is and reflects those people's opinions rather than London Bilingualism's editorial stance.
We strongly advise all readers to consult a qualified medical professional before acting on any medical, health, dietary, or pharmaceutical information found on this website. Since every person's health situation is different, only a qualified healthcare provider who is familiar with your medical history can offer you advice that is suitable for you.
In a similar vein, any legal, regulatory, or compliance-related information found on this platform is provided solely for informational purposes and should not be used without first obtaining independent legal counsel from a licensed attorney.
You understand and agree that London Bilingualism, its editors, contributors, and affiliated parties are not responsible for any decisions made using the information on this website.
