Can AI help us save endangered languages?

Around half of the world’s languages face extinction, according to UNESCO, primarily due to parents abandoning them in favor of more widely spoken tongues, schools dropping them from curricula, or governments failing to recognize their official status. Campaigners argue that AI could play a crucial role in preserving these languages, along with the cultural heritage they carry, though critics warn the technology may worsen inequality by prioritizing only a handful of dominant languages. Sophia Smith Galer, a journalist and author, explains that endangered languages often disappear when younger generations no longer learn them, erasing centuries of history and tradition. She highlights AI as a potential tool to document, translate, and revive these languages, but notes that its current training data overwhelmingly favors major languages like English, Spanish, or Mandarin, risking further marginalization of lesser-spoken ones. In the U.S., Ivory Yang developed an AI tool specifically to preserve Nüshu, an endangered language traditionally used by women in her grandmother’s region. The tool digitizes spoken Nüshu, translates it, and even generates new phrases, offering a lifeline to a language that has struggled to survive modernization. Yang’s project demonstrates how AI can bridge gaps in documentation and accessibility, but also raises questions about sustainability—whether such tools can thrive without continuous funding or community engagement. Experts caution that AI-driven preservation efforts must address broader systemic issues, such as lack of political will and limited resources for minority languages. Without these, even the most advanced AI tools risk becoming temporary solutions rather than long-term safeguards. The debate underscores a tension: while AI offers unprecedented opportunities to save endangered languages, its unequal development could deepen the very disparities it aims to overcome.