A 2017 tale, akin to a cautionary tale, continues to circulate among linguists and AI researchers. In the West Bank, a Palestinian man wrote “good morning” in Arabic on Facebook. According to Facebook’s AI, it means “attack them.” He was taken into custody. When the mistranslation was eventually discovered, the harm had already materialized in a very tangible and physical manner. This type of mistake raises questions about how much faith we should have in systems that don’t truly comprehend what they’re reading.
Since then, AI translation has significantly improved. Exchanges at the tourist level are handled with remarkable fluency by programs like Google Translate and Apple’s live interpretation capabilities. Asking for directions in Lisbon, placing an order for coffee in Osaka, and perusing a menu in Istanbul are all largely resolved. However, the technology begins to appear much less capable when you sit down at a family dinner in Karachi where Urdu and English blend together in the middle of a sentence, where a grandmother’s silence is more powerful than her words. Conversations that are bilingual involve more than just two languages operating simultaneously. No algorithm has been trained to perceive them; they are emotional, multi-layered, and full of code-switching and implicit meaning.
According to research from 2025, standard machine translation loses almost half of all contextual meaning. Anyone should be concerned by that figure alone, but in high-stakes situations like medical consultations, court cases, and diplomatic negotiations, it becomes even more concerning. When a doctor in a London hospital uses AI to give medication instructions to a patient who speaks Bengali, they are not only running the risk of using clumsy language. There is a very narrow margin of error. Nevertheless, the trend toward AI-first communication tactics keeps picking up speed, sometimes with little regard for what’s being left behind.
There is a structural component to the issue. Due to their training on English-language data, the majority of large language models have cultural defaults that are skewed toward Western norms. Idiomatic phrases become flattened. Humor fades. When translated by a system that hasn’t assimilated the cultural shorthand, the French expression “tomber dans les pommes”—literally, “to fall into the apples,” meaning to faint—becomes meaningless. Machines that still parse sentences for literal content hardly notice sarcasm, which functions differently in Arabic, Japanese, and Spanish. A system that can process millions of words per second but is unable to recognize humor has an almost poignant quality.
AI finds bilingual conversations especially challenging because they don’t adhere to strict rules. In a Spanglish conversation, friends in Los Angeles deliberately switch between languages for emphasis, intimacy, humor, and occasionally just habit. The switch itself has significance. Saying “no manches” rather than “no way” indicates identity, intimacy, and a shared cultural understanding that is unique to the speakers. AI is bilingual. One conversation is seen by bilingual speakers.
Government-backed IndiaBy creating foundational models that are in line with the nation’s remarkable linguistic diversity—which encompasses dozens of languages and hundreds of dialects—AI Mission is attempting to address some of this. It’s an ambitious endeavor that recognizes that culturally intelligent AI needs different data, collected with consideration for social context and regional expression, rather than just better data. It’s unclear if it will be successful on a large scale, but at least the idea seems sound.
It’s easy to present this as a transient issue that will eventually be resolved by larger models and better training data. Perhaps they will. However, language is more than just a means of communicating. The way a grandmother’s proverb lands differently from its dictionary definition is an example of identity and belonging. No dataset can fully capture the emotional memory, social awareness, and years of lived experience that bilingual speakers use to naturally navigate these layers. Travelers purchasing train tickets may be able to overcome the language barrier thanks to AI earbuds, which is really helpful. However, machines continue to stand outside the door, listening without fully comprehending what they are hearing, during the conversations that truly matter—the ones where respect is expressed, trust is established, and meaning exists between the words.
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.
