29 Jun 2026 Language loss and AI

Will artificial intelligence (AI) help or hinder the survival of minority languages into the next century?

According to Ethnologue, a database of the world’s 7,168 living languages, almost half are endangered. Over 88 million people speak a language that may not survive this century except in historical records.

Oceania has 733 endangered languages, of which Papua New Guinea – with a population of just 8.8 million – accounts for 312. At the country level, Indonesia leads with 425 endangered languages – a direct result of a national policy that made Bahasa Indonesia the sole medium of education and government in more than 700 distinct linguistic communities.

Africa records 428 endangered languages, mainly clustered around equatorial zones where conflict, drought, and displacement are destroying community life. North and Central America account for 222, with the United States contributing 180.

Language loss is aggravated by several factors: economic pressures that often force young people to adopt dominant languages for employment, education, and social mobility; cultural assimilation and education policies that prioritize national languages; urbanization and migration; forced assimilation and political suppression of Indigenous languages; and globalization and modernization, which prefer global languages in business, media, and the development of new technologies.

The role of artificial intelligence is controversial. In “AI will make language barriers disappear – and diminish our understanding of other cultures” (The Guardian, 9 May 2026), Italian novelist and interpreter Diego Marani warns of a “frontier in artificial intelligence and in the realm of language from which there will be no turning back.” Noting that machines will perform translation and interpretation tasks quickly, efficiently, and economically, he points to a profound change in human communication – akin to the 6,000 year-old transition from speech to writing.

“The first effect of the AI translation revolution will be to render the study and learning of languages superfluous for individuals. It will be enough to turn to our phones to understand whoever speaks to us and to translate our own speech into any language. Eventually, we shall be able to read information in every language, to write texts that can be read from one end of the world to the other. Yet knowledge – true understanding of others, of their cultures and customs, of the cast of mind of another country – will not thereby become ours. This body of knowledge will reside in AI systems, not in us.”

AI is a prime example of Arthur C. Clarke’s third law, which states that “Any sufficiently advanced technology is indistinguishable from magic.” But like magic, AI is all smoke and mirrors. It creates seemingly grammatically correct but sometimes factually wrong, “creative” translations, or invents information, particularly in less popular languages. It struggles with irony, sarcasm, implied meanings, and highly contextual cultural cues. AI models predominantly trained on English reflect Western-centric views, treat dialects as less “correct” and often fail to understand regional variants. And over-dependence on AI translation can diminish people’s critical thinking and willingness to learn another language.

In practical terms, these deficits can be overcome by developing models trained on a more diverse range of languages; by including native speakers of minority languages in the development and validation of AI tools; and by creating AI models that cater to regional languages and contexts.

But if we are to avoid language loss, the bottom line is not to allow the illusionary magic of machines to replace human empathy and intuition. In her 1998 essay, specialist in American Indian languages Marianne Mithun wrote, “The loss of languages is tragic precisely because they are not interchangeable… precisely because they represent the distillation of the thoughts and communication of people over their entire history.”

Despite its obvious usefulness, AI imperils languages. Technological determinism should not be allowed to undermine what Diego Marani identifies as “true understanding of others, of their cultures and customs.” Nor to erase the defining characteristic of all human beings: their languages.

Graphic: The Global Education Project  www.theglobaleducationproject.org