Idea 18 for 2025: AI, Language Learning and the Hope for Linguistic Preservation

To me, and for many, language is tied to culture, a learned and shared symbolic meaning system through we think and communicate.

With some 7000+ languages in the world, languages are an emblem of immense cultural diversities of humans.

Yet, with the growing advent of AI, and the refinement of language translation from the familiar Google Translate to now increasingly live translation, do we need to learn languages anymore?

This thought crosses my mind as I consider experiences from Study Abroad as well as travel experiences, where I see tourists turn increasingly to their phones and other devices to quickly translating their  thoughts to local recipients, seemingly removing a key barrier of an intercultural encounter. Indeed, this past weekend at a BBQ, I encountered a fellow guest who was sporting a pair of Meta AI Ray-Bans, that could take and upload photos and video and offer, the guest said, narration of what you were seeing with no need to look things up on your phone.

A recent episode of CBC’s Sunday magazine probed the question further, noting that with the digital and machine learning translation tools, enrolment in language learning classes was experiencing decline. It seems that people are deciding that there is no longer a need to learn languages.

Is it so simple?

Well, CBC host interviewed various specialists who talked about their experiences “lost in translation” or mistranslation. While language teachers can share their specific biases based on how and in what context they have learned and taught a language, a LLM is not doing the same thing; that is, we don’t know these large companies are being trained, and what sources of language they are using.

One host gave the example of learning Mandarin and using AI tools that, upon further investigation, had been trained on Chinese state media, which favoured for formal and respectful language that may not accurately represent everyday speech.

Yet another guest, an academic talked about the sounds of language that can be lost and eroded in machine-translation, given examples between phrases between the linguistically diverse English vs. Arabic.

This guest also spoke about the fact that “technology always has an agenda,” and that while the languages with the greatest number of speakers may be sought after by AI companies, what about the plurality of the 7000+ languages with smaller speech communities across the globe?

Indeed, the guest suggested that by working with local communities, AI companies could help preserve these languages and with that, their symbolic meanings, cultural worldviews, unique sounds.

And, oh, there was one language that seemed to be bucking the trend and growing in popularity to learn as an additional language, fuelled by the globalization of its various media. Can you guess which one?

Before you ask AI, here’s the answer: Korean. 😊

Two people speak via Google's Meet live translation service, during a demonstration at Google's I/O event in May. (Google)