AI uses up to 180 times more energy for some languages

AI models may use 1.4 to 180 times more energy to generate text in languages other than English, depending on the language, a recent study found. ERR reported the finding in a technology commentary by Kristjan Port on R2.

The origins of language remain disputed. One theory holds that language emerged only once in human evolution, meaning all modern spoken languages are linked in some way. Some researchers have suggested it began among early humans in southwestern Africa.

One hypothesis is that click sounds, used in languages spoken by peoples in southwestern Africa, helped people distinguish more things than the available vocal sounds could name. These sounds, made with the tongue and palate, carried information even as natural selection favoured changes to the face and throat that enabled a wider range of vocal sounds and new word-forming sounds.

These are only some of many theories. The precise place and time where language began are unknown, and researchers disagree on whether language is a biological programme or a tool created by people. The most likely view, Port writes, combines the two: evolution gave humans the capacity to acquire language, while culture shaped languages into their present forms.

The development of Nicaraguan Sign Language offers an example of how language can evolve. During a literacy campaign launched by Nicaragua’s government in the 1980s, a group of deaf children who had no common language were brought together. Nicaragua had no sign language at the time. Within a few years, the children had created a functioning language with its own grammar, which their teachers initially could not understand.

The experience supported the idea that the basic structure for language is present in the brain, need not be spoken, and can develop through social interaction.

As of 2026, about 7,500 spoken and signed languages remain, and roughly half are classified as endangered. That is only about a tenth of all the languages that have existed. Social, economic and political pressures are the main drivers of language loss, leading communities to give up their mother tongues in favour of a dominant language.