The title is the original title from the Guardian article, but does not match the content accurately. The fundamental problem is that Icelandic is a small language, and English-language media is already dominating. This puts Icelandic in danger of dying. Now, AI threatens to accelerate that process, especially since of all things the AI company Anthropic is now working with teachers to provide teaching material for them:
“Having this language that is spoken by so very few, I feel that we carry a huge responsibility to actually preserve that. I do not personally think we are doing enough to do that,” she said, not least because young people in Iceland “are absolutely surrounded by material in English, on social media and other media”.
Katrín has said that Iceland has been “quite proactive” in pushing for AI to be usable in Icelandic. Earlier this month, Anthropic announced a partnership with Iceland’s ministry of education, one of the world’s first national AI education pilots. The partnership is a nationwide pilot across Iceland – giving hundreds of teachers across Iceland access to AI tools.
This is nonsense. The more connected the world, the fewer languages we need to communicate. It has nothing to do with AI and has been going on for decades:
https://en.wikipedia.org/wiki/Language_death
People are also losing their accents because of the Internet:
https://en.wikipedia.org/wiki/Dialect_levelling
BTW: Technically it started before the Internet... Caused by migration and globalization. However, the Internet has accelerated it to the point where parents are noticing it in their children (single generation loss of accents/dialect).
As I understand the article, the concern is that AI will accelerate that even more - because small languages have little training data for LLMs.
To be fair to the former PM and author, all of these points are touched upon one way or another in the article and so-called AI is merely another layer of threat to the existence of such a small language community. She didn't consider Icelandic safe before the models popped up and just now got worried. And while it's probably too early to have scientific proof about the influence of models to back up her argument, I don't think it's nonsense to think that way. And I think the guardian headline is misleading.