Indigenous youth use AI to preserve languages disappearing every 2 weeks

Young Indigenous technologists are developing AI-powered tools to help preserve and revitalize endangered languages in their communities, addressing a crisis where the UN estimates an Indigenous language dies every two weeks. These projects, led by individuals like Danielle Boyer with her SkoBot language-learning robot and Jared Coleman’s AI-powered Paiute translator, represent a growing movement to reverse generational language loss while maintaining ethical control over cultural resources.

The big picture: Indigenous communities worldwide are racing against time to preserve languages that colonization systematically suppressed, with half of all global languages projected to disappear by 2100.

How the technology works: Boyer’s SkoBot uses AI speech recognition to translate English words into Anishinaabemowin through pre-recorded audio files featuring children’s voices from her community.
• The coffee mug-sized robot sits on a user’s shoulder and responds to English words with their Indigenous language equivalents—saying “Boozhoo” when users say “hello.”
• Students build the robots themselves in classrooms, combining STEM education with language learning through hands-on assembly and programming.
• Coleman’s system uses OpenAI’s GPT-3.5-turbo and GPT-4 models trained on Owens Valley Paiute words to create basic sentences and power an online dictionary and translator.

Why this matters: Language loss represents cultural extinction, as Boyer explains: “When you lose your language, you lose such a key component of your culture and your ways.”
• Many Indigenous communities experienced rapid generational decline, with grandparents speaking fluently while their grandchildren know only fragments.
• The COVID-19 pandemic accelerated this crisis by claiming many elder speakers who served as primary language repositories.

Ethical considerations: These technologists are deliberately avoiding AI practices that could exploit Indigenous cultural resources.
• Boyer chose pre-recorded human voices over AI-generated audio because “languages are living things” and learning “should always happen with a community member.”
• Community members retain ownership of their voice recordings under an ethical AI framework developed specifically for the project.
• Coleman avoids uploading verbatim elder recordings to prevent potential misuse by AI companies, particularly of sacred songs and stories.

What they’re saying: “The last person to speak the language fluently in my family was my great grandfather,” Coleman told CNN, explaining how boarding schools prohibited Indigenous languages.
• “My great grandpa went to a boarding school where it was prohibited to speak the language, so my grandma didn’t get taught the language. That’s the sad history of the language in my family, and it’s the same for a lot of people in my tribe and in a lot of other tribes.”
• Boyer emphasizes documentation importance: “making sure that my language is recorded and well-documented but in a way that’s not being exploited by companies that are not from our communities.”

Accuracy concerns: Both developers have witnessed mainstream AI chatbots producing incorrect representations of their languages, which could perpetuate cultural misunderstandings.
• “Language is so much more than just its words,” Coleman noted. “It encodes an entire culture and an entire history along with it.”