Many languages have few speakers and little digital material, which puts them at risk and leaves them unsupported by technology. AI offers help: with community involvement, models can assist translation, transcription and learning tools even for languages with limited data.
This can support speakers, aid documentation and make devices more usable in a mother tongue. The difficulty is that models need data these languages lack, and doing this respectfully requires working with the communities themselves rather than extracting from them. Handled that way, AI becomes a tool that helps keep linguistic diversity alive instead of accelerating its loss.