OtuhoAI

About Otuho AI

Otuho has a rich oral tradition, but very little of it is written online, so today's AI systems barely understand it. An AI learns a language from a large number of examples written by the people who speak it. This project gathers those examples, with the community and for the community, so that an open AI model can learn to understand, write and speak Otuho.

What we are building

We don't start from nothing. We take an existing open-source model and teach it Otuho, using texts that are already published, such as the Otuho Bible, together with new writing from speakers on this site. The goal is an assistant you can talk to in Otuho, by text or by voice, that answers in fluent Otuho.

Read the full plan

Privacy and consent

Every contributor agrees to the following before contributing:

  • What I write here will be used to teach a computer model to understand, write and speak Otuho.
  • My name and phone number stay private. They are never published or shared with the data.
  • My contributions may be shared with other researchers only if the Otuho community's data governance group approves.
  • I will not share sacred or restricted knowledge here unless I have the right to.
  • I can ask for my contributions to be deleted at any time.

Who owns the data

The Otuho community. We follow the CARE principles for Indigenous data: collective benefit, authority to control, responsibility and ethics. A community data governance group decides how the data is used and shared. All of the code is open source.